‘

Double Integrals

Suppose the function f is continuous on R , the rectangle a ≤ x ≤ b , c ≤ y ≤ d . If ( uij , v ij ) is any point in

the ij -th subrectangle, we define the definite integral of f over R .

∫ fdA= Δ x lim
, Δ y →0
∑ f ( uij , v ij ) Δ x Δ y
R i,j

Such an integral is a called a double integral.

Sometimes we think of dA as being the area of an infinitesimal rectangle of length dx and height dy , so
that dA=dxdy . Then we use the notation

∫ fdA=∫ f ( x , y ) dxdy
R R

The case when R is not rectangular is separately considered. Sometimes we think of dA as being the
area of an infinitesimal rectangle of length dx and height dy , so that dA=dxdy . The we use the
notation

∫ fdA=∫ f ( x , y ) dxdy ,
R R

Integration of the Double Integral as Volume

Just at the definite integral of a positive one-variable function can be interpreted as an area, so the
double integral of a positive two-variable function can be interpreted as a volume. In the one-variable
case we visualize the Riemann sums as the total area of rectangles above the subdivisions. In the two-
variable case we get solid bars instead of rectangles. As the number of subdivisions grows, the tops of
the bars approximate the surface better, and the volume of the bars gets closer to the volume under the
graph of a function.

If x , y , z represent length and f is positive, then

Volume under f above region R=∫ fdA

R

Let R be the rectange 0 ≤ x ≤ 1 and 0 ≤ y ≤ 1. Use the Riemann sums to make upper and lower estimates
2 2
of the volume of the region above R and under the graph of z=e− x + y .

Interpretation of the Double Integral as Area

In the special case that f ( x , y )=1 for all points ( x , y ) in the region R , each term in the Riemann sum is
of the form 1. Δ A=Δ A and the double integral gives the area of the region R :

Area ( R )=∫ 1 dA=∫ dA

R R

Interpretation of the Double Integral as Average Value

As in the one-variable case, the definite integral can be used to defined the average value of a function:
 1
Average value of f on the region R = ∫ fdA
Area of R R

We can rewrite this as

Average value x Area of R = ∫ fdA

R

If we interpret the integral as the volume under the graph of f , then we can think of the average value
of f as the height of the box that is on the same base. Imagine that the volume under the graph is made
out of wax. If the was melted within the perimeter of R , then it would end up box-shaped with height
equal to the average value of f .

L
Area=∫ f ( x ) dx
0

L
1
Average Va lue= ∫ f ( x ) dx
L0
L
L × Average Value=∫ f ( x ) dx
0

Integral over Regions that Are Not Rectangles

We defined the definite integral ∫ f ( x , y ) dA , for a rectangular region R . Now we extend the definition
R
to regions of other shapes, including triangles, circles, and regions bounded by the graph of the
piecewise continuous functions.

To approximate the definite integral over a region R , which is not rectangular, we use a grid of
rectangles approximating the region. We obtain this grid by enclosing R in a large rectangle and
subdividing that rectangle; we consider just the sub-rectangles which are inside R .

As before, we pick a point ( uij , v ij ) in each sub-rectangle and form a Riemann sum

∑ f ( uij , v ij) Δ x Δ y
i, j

This time, however, the sum is over only those sub-rectangles within R . For example, in the case of the
fox population we can use the rectangles which are entirely on land. As the subdivisions become finer,
the grid approximates the region R more closely. For a function, f , which is continuous on R , we define
the definite integral as follows:

∫ fdA= Δ x lim
, Δ y →0
∑ f ( uij , v ij ) Δ x Δ y
R i,j

where the Riemann sum is taken over the sub-rectangeles inside R .

You may wonder why we can leave out the rectangles which cover the edge of R – if we included them,
might we get a different value for the integral? The answer is that for any region that we are likely to
meet, the area of the sub-rectangles covering the edge tents to 0 as the grid becomes finer. Therefore,
omitting these rectangles does not affect the limit.

Convergence of Upper and Lower Sums to Same Limit

We have said that if f is continuous on the rectangle R , then the difference between upper and lower
sums for f converges to 0 as Δ x and Δ y approach 0. In the following example, we show this in a
particular case. The ideas in this example can be used in a general proof.

Let f ( x , y )=x 2 y and let R be the rectangle 0 ≤ x ≤ 1 ,0 ≤ y ≤ 1. Show hat the difference between upper
and lower Riemann sums for f on R converges to 0, as Δ x and Δ y approach 0.

The difference between the sum is

∑ M ij Δ x Δ y −∑ Lij Δ x Δ y =∑ ( M ij −Lij ) Δ x Δ y ,

where M ij and Lij are the maximum and minimum of f on the ij -th sub-rectangle. Since f increase in
both the x and y directions, M ij occurs at the corner of the sub-rectangle farthest from the origin and
Lij at the closet. Moreover, since the slopes in the x and y directions don’t decrease as x and y
increase, the difference M ij −Lij is the largest in the sub-rectangle Rnm which is farthest from the origin.

Thus,

∑ ( M ij−Lij ) Δ x Δ y ≤ ( M nm−L nm) ∑ Δ x Δ y=( M nm−Lnm ) Area ( R ) .

Thus, the difference converges to 0 as long as ( M nm−Lnm ) does.

( ) (1− m1 )
2
2 1
M nm−Lnm=( 1 ) ( 1 )− 1−
n
2 1 1 2 1
¿ − 2+ − +
n n m nm n2 m

The right-hand side converges to 0 as n , m→ ∞ , that is, as Δ x , Δ y →0 .

Table 16.7 gives values of f ( x , y ), the depth of volcanic ash, in meters, after an eruption. If x and y are
in kilometers and R . If x and y are in kilometers and R is the rectangle 0 ≤ x ≤ 100 , 0< y ≤100,
estimate the volume of volcanic ash in R in km 3.

Setting up a Double Integral and Approximating it by Double Sums

Consider the function z=f ( x , y )=3 x 2− y over the rectangular region R=[ 0 ,2 ] ×[0 , 2].
 a. Set up a double integral for finding the value of the signed volume of the solid S that lies above
R and “under” the graph of f .
b. Divide R into four squares m=n=2, and choose the sample point as the upper right corner
point of each square ( 1 ,1 ) , ( 2 ,1 ) , ( 1 ,2 ) , and (2 , 2) to approximate the signed volume of the
solid S that lies above R and “under” the graph of f .
c. Divide R into four squares with m=n=2, and choose the sample point as the midpoint of each

square: ( 12 , 12 ) ,( 32 , 12 ), ( 12 , 32 ) , and ( 32 , 32 ) to approximate the signed volume.

Solution

a. As we can see, the function z=f ( x , y )=3 x 2− y is above he plane. To find the signed volume of
S, we need to divide the region R into small rectangles Rij , each with area Δ A and with sides
¿ ¿
Δ x and Δ y , and choose ( x ij , y ij ) as sample points in each Rij . Hence, a double integral is set up
as
m n
V =∬ ( 3 x − y ) dA= lim
2

m , n→ ∞
∑ ∑ [
i=1 j=1
3 ( x ¿ij ) − y ¿ij ] Δ A
2

R
b. Approximating the signed volume using a Riemann sum with m=n=2 we have
Δ A=Δ x Δ y =1× 1=1. Also, the sample points are (1,1), (2,1), (1,2), and (2,2).
2 2
V =∑ ∑ f ( x ¿ij , y ¿ij ) Δ A
i=1 j =1
c. Approximating the signed volume using a Riemann sum with m=n=2, we have
Δ A=Δ x Δ y =1× 1=1.
2 2
V =∑ ∑ f ( x ¿ij , y ¿ij ) Δ A
i=1 j =1

ITERATED INTEGRALS

In this section we see how to compute double integrals exactly using one-variable integrals.

The Fox Population Again: Expressing a Double Integral as an Iterated Integral

To estimate the fox population, we computed a sum of the form

Total population ≈ ∑ f ( u ij , v ij ) Δ x Δ y
i, j

where 1 ≤i ≤ n and 1 ≤ j≤ m and f ( u ij , v ij ) is the function evaluated at the point ( uij , v ij ).

(∑ )
m n
Total Population ≈ ∑ f ( u ij , v ij ) Δ x Δ y
j=1 i=1

n 180
The inner sum, ∑ f ( uij , v ij ) Δ x , approximates the integral ∫ f ( x , v ij ) dx. Thus, we have
i=1 0
 (∫ )
m 180

Total population ≈ ∑ f ( x , v ij ) dx Δ y
j=1 0

180
The outer Riemann sum approximates another integral, this time with integrand ∫ f ( x , y ) dx which is a
0
function of y . Thus, we can write the total population in terms of nested, or iterated, one-variable
integrals:
150
Total population = ∫ ¿ ¿.
0

Since the total population is represented by ∫ fdA , this suggests the method computing double
R
integrals in the following theorem.

3 2

∑ ∑ ( xij + y ij )
i=1 j=1

3
¿ ∑ ( x i 1+ y i 1+ x i 2 + y i 2 )
i=1

¿ ( x 11 + y 11 ) + ( x 12+ y 12 ) + ( x 21+ y 21) + ( x 22 + y 22 )+ ( x 31+ y 31 ) + ( x 32+ y32 )

Theorem 16.1: Writing a Double Integra as an Iterated Integral

If R is a rectangle a ≤ x ≤ b , c ≤ y ≤ d and f is a continuous function on R , then the integral of f over R

exists and is equal to the iterated integral

(∫ )
y=d x=b

∫ fdA= ∫ f ( x , y ) dx dy
R y=c x=a

( )
y=d x=b d b
The expression ∫ ∫ f ( x , y ) dx dy can be written as ∫ ∫ f ( x , y ) dxdy .
y=c x=a c a

A building is 8 meters wide and 16 meters long. It has a flat roof that is 12 meters high at one corner and
10 meters high at each of the adjacent corners. What is the volume of the building?
⃗
A=⟨ 0 ,0 ,12 ⟩
⃗
B=⟨ 16 , 0 , 10 ⟩
⃗
C =⟨ 0 , 8 ,10 ⟩
⃗
AB= ⟨ 16 , 0 , 10 ⟩ −⟨ 0 , 0 , 12 ⟩ =⟨ 16 ,0 ,−2 ⟩
 ⃗
AC=⟨ 0 , 8 ,10 ⟩− ⟨ 0 , 0 ,12 ⟩= ⟨ 0 , 8 ,−2 ⟩

n⃗ =⃗
AB × ⃗
AC

| i^ ^j k^
¿ 16 0 −2
0 8 −2
|
¿ i^ ( 0−(−16 ) )− ^j (−32−0 ) + k^ (128−0)
^
¿ 16 i+32 ^j+ 128 k^

⃗
AP= ⟨ x , y , z−12 ⟩
⃗
AP . ⃗n= ⟨ x , y , z −12 ⟩ . ⟨ 16 , 32,128 ⟩ =0
16 x +32 y +128 z−12 ×128=0
128 z=12 ×128−16 x −32 y
1 1
z=12− x− y
8 4
The Order of Integration

In computing the fox population, we could have chosen to add columns (fixed x ) firs, instead of the
rows.

( )
b d

∫ f ( x , y ) dA=∫ ∫ f ( x , y ) dy ¿¿ dx
R a c

where R is the rectangle a ≤ x ≤ b and c ≤ y ≤ d .

For any function we are likely to meet, it does not matter in which order we integrate over a rectangular
region R ; we get the same value for the double integral either way.

( ) (∫ )
d b b d

∫ fdA=∫ ∫ f ( x , y ) dx dy=∫ f ( x , y ) dy dx
R c a a c

Iterated Integrals Over Non-Rectangular Regions

The density at the point ( x , y ) of a triangular metal plate is shown in Figure 16.13, is δ ( x , y ). Express its
mass as an iterated integral.

Mass of a rectangle ≈ Density. Area ≈ δ ( x , y ) Δ x Δ y

Mass ¿ ∫ δ ( x , y ) dA
R

a≤ x≤b,c ≤ y ≤d
 b d

∫∫ f ( x , y ) xdy
a c

2−2 x

∫ δ ( x , y ) dy
0

1 2−2 x
Mass=∫ ∫ δ ( x , y ) dxdy
0 0

1
1− y
2 2

Mass=∫ ∫ δ ( x , y ) dxdy
0 0

Limits on Iterated Integrals

 The limits on the outer integral must be constants.

 The limits on the inner integral can involve only the variable in the outer integral. For example, if
the inner integral is with respect to x , its limits can be functions of y .
6 2

∫∫ x √ y3 +1 dydx
0 x
3

x
y= ⇒ x=3 y
3
y=2, x=0
2 3y

∫∫ x √ y 3+1 dxdy
0 0

( )
2 2 x=3 y
x
¿∫
2
√ y +13
dy
0 x=0

2 2 1
9y 3
¿∫ ( y +1 ) 2 dy
0
2

[ ] =27−1=26
3 2

¿ ( y +1 )
3 2
0

1
9 y2 3
∫ 2 ( y +1 ) 2 dy
1
9 y 2 2 du
¿∫ ( u)
2 3y
2
 1
3
¿ ∫ u 2 du
2
3
3 2
¿ × u 2+ c
2 3
3
Let u= y + 1
2
du=3 y dy
du
dy = 2
3y