Chapter 4

Higher integration

Having been successful in generalizing differentiation to multivariable func-

tions, we now turn our attention to integration. We start with the case of
two-variable functions.

4.1 Double integrals

4.1.1 Motivation
Problem 1 (Finding mass from a density function).
Suppose a two-dimensional metal plate (represented by a closed and
bounded region of the xy-plane) is constructed from various material, and
has non-uniform density. In other words, there is a non-constant density
function d(x, y) which gives the density of the plate at the point (x, y). We
want to know: what is the total mass of the plate?
The next motivating problem indicates a good visual guide to defining
integration of a function f (x, y) over a two dimensional region R.
Problem 2 (Volume of the solid underneath y = f (x, y) over a region of the
xy-plane).
Given a function f (x, y), the graph z = f (x, y) is a surface. Suppose
f (x, y) ≥ 0 on a closed bounded region R. Then there is a 3-dimensional
solid S bounded between the portion of the surface above R and the region
R in the xy-plane. In other words, S consists of all the points (x, y, z), where
(x, y) is in R, and 0 ≤ z ≤ f (x, y). We want to know: what is the volume of
S?

1
CHAPTER 4. HIGHER INTEGRATION 2

To answer these questions, we will need to define the integral of a function

of two variables over a closed and bounded region of the plane. As is often
so helpful, let’s start by recalling the approach we used with single variable
functions.
• Suppose f (x) is a continuous function on an interval [a, b].
b−a
• First, partition [a, b] into n subintervals of equal length ∆x =
n
and denote the endpoints of these subintervals by x0 < x1 < · · · < xn ,
where x0 = a, xn = b.
• Pick a sample point x∗i in each subinterval [xi−1 , xi ]. This choice may
be either arbitrary (to define the general Riemann sum) or specific
choice eg. left/right endpoint or midpoint (which are useful for defining
specific Riemann sum approximation methods).
n
X
• Consider the (general) Riemann sum f (x∗i )∆x
i=1

• If the limit of the Riemann sums exists, as the partition is made finer
and finer, it is defined to be the value of the definite integral. In other
words, we define
Z b Xn
f (x)dx = lim f (x∗i )∆x
a n→∞
i=1

if the limit exists. Notice that this equivalent to defining

Z b Xn
f (x)dx = lim f (x∗i )∆x
a ∆x→0
i=1

since, if the length of the subintervals goes to 0, the number of subin-

tervals must go to infinity.
We can follow a similar plan for defining the integral of a continuous
function on a (closed and bounded) region of the plane. When it comes to
finding computational method for these double integrals, we will later see
that it will be possible to reduce to integrals of single variable functions (for
which previous computational methods can be used). First, we need to define
appropriate Riemann sums for functions of two variables.
Rather than try to deal with arbitrary closed and bounded regions im-
mediately, we’ll first work with simpler case where the region is a rectangle.
CHAPTER 4. HIGHER INTEGRATION 3

4.1.2 Double Riemann sums on a rectangle

• Suppose a function f (x, y) is defined on a closed rectangle R consisting
of all points (x, y) with a ≤ x ≤ b and c ≤ d. We can denote such a
region by: (
a≤x≤b
R:
c≤y≤d

or R : [a, b] × [c, d].

• Partition [a, b] into m equal-sized subintervals [x0 , x1 ], [x1 , x2 ], . . . , [xm−1 , xn ].

b−a
In other words, let x0 = a, xm = b and xi = a+i·∆x, where ∆x =
n
is the length of each subinterval [xi−1 , xi ].

• Partition [c, d] into n equal-sized subintervals [y0 , y1 ], [y1 , y2 ], . . . , [yn−1 , ym ].

d−c
In other words, let y0 = c, yn = d and yj = c+j ·∆y, where ∆y =
m
is the length of each subinterval [yj−1 , yj ].

• The above partitions of segments on the x and y axes induce a partition

of the rectangle R into sub-rectangles Rij where
(
xi−1 ≤ x ≤ xi
Rij :
yj−1 ≤ y ≤ yj

Notice that the area of Rij is ∆x∆y = ∆A and the sum of areas of all
the rectangles Rij is equal to area of R.
CHAPTER 4. HIGHER INTEGRATION 4

• In each rectangle Rij choose a sample point (x∗ij , yij∗ ). Again, for the
purposes of defining the general double Riemann sum, we can choose
arbitrary (x∗ij , yij∗ ), and if we want to define a specific double Riemann
sum for approximation, we can make a specific choice, e.g. upper right
corner: (x∗ij , yij∗ ) = (xi , yj ).

• A double Riemann sum for the partition is defined as:

m X
X n
f (x∗ij , yij∗ ) ∆A
i=1 j=1

which can be read as

m n
!
X X
f (x∗ij , yij∗ ) ∆A
i=1 j=1

or equivalently as !
n
X m
X
f (x∗ij , yij∗ ) ∆A
j=1 i=1

Notice that this double sum can be rewritten as a single sum

N
X
f (x∗k , yk∗ ) ∆A
k=1

where N = m · n, and the index k corresponds to some order of counting the

N sub-rectangles of the partition. For example, the double sum
m n
!
X X
f (x∗ij , yij∗ ) ∆A
i=1 j=1

corresponds to counting up each column (increasing j) before moving to the

N
X
next column (increasing i). We still call f (x∗k , yk∗ ) ∆A a double Riemann
k=1
sum, since it counts over two dimensional regions.
We can visualize a double Riemann sum as the sum of volumes of boxes
approximating the solid under a surface.
CHAPTER 4. HIGHER INTEGRATION 5

Fact 4.1.1. If f (x, y) ≥ 0 on a rectangle R then the each double Riemann

sum is an approximation of the volume of the solid with base R and bounded
above by the graph of f .
Example 4.1.1. Estimate the volume of the solid that lies above the square
[0, 2] × [0, 2] and below the graph of the function f (x, y) = 16 − x2 − 2y 2 .
Solution. Let us choose n = m = 2, that is, we partition both the x and
y intervals each into two subintervals. In this case ∆x = 1, ∆y = 1. As a
result we obtain a partition of R into four smaller rectangles:

R1,1 = [0, 1]×[0, 1], R1,2 = [0, 1]×[1, 2], R2,1 = [1, 2]×[0, 1], R2,2 = [1, 2]×[1, 2]

Next, choose sample points:

(x∗1,1 , y1,1
∗
) = (0, 1), (x∗1,2 , y1,2
∗
) = (0, 2), (x∗2,1 , y2,1
∗
) = (1, 1), (x∗2,2 , y2,2
∗
) = (1, 2)

which are the upper left corners of the smaller rectangles. Hence, we obtain
the double Riemann sum giving the required estimation since f is positive
on R
2 X
X 2
f (x∗ij , yij∗ ) ∆A = f (0, 1) · 1 · 1 + f (0, 2) · 1 · 1 + f (1, 1) · 1 · 1 + f (1, 2) · 1 · 1
i=1 j=1

= 14 + 8 + 13 + 7 = 42
We are often especially interested in the following double Riemann sums.
CHAPTER 4. HIGHER INTEGRATION 6

Definition 4.1.2 (Upper/lower double Riemann sums).

If the sample points (x∗ij , yij∗ ) are each picked so that each f (x∗ij , yij∗ ) is
the maximum value of f on Rij , then call the corresponding double Riemann
sum an upper double Riemann sum.
Similarly, if the sample points are picked so that each f (x∗ij , yij∗ ) is the
minimum value of f on Rij , then call the corresponding double Riemann sum
an lower double Riemann sum.
Example 4.1.2. Given the following contour diagram for z = f (x, y)

Let R = [−2, 4] × [−4, 2]. Find the upper and lower double Riemann sums
for the partition of R with ∆x = ∆y = 2.
Solution. We can determine the maximum and minimum values of f in each
subrectangle from the contour diagram. For example, the minimum value of
f on [−2, 0] × [−2, 0] is −6, occurring at (−2, −2), and the maximum is 0,
occurring at (0, 0).
So the lower double Riemann sum is:

(−6)(2)(2) + (−4)(2)(2) + (−2)(2)(2)

+(−6(2)(2) + (−4)(2)(2) + (−2)(2)(2)
+(−18)(2)(2) + (−16)(2)(2) + (−14)(2)(2) = −288
CHAPTER 4. HIGHER INTEGRATION 7

and the upper double Riemann sum is:

(0)(2)(2) + (2)(2)(2) + (4)(2)(2)
+(0)(2)(2) + (2)(2)(2) + (4)(2)(2)
+((−4)(2)(2) + (−2)(2)(2) + (0)(2)(2) = 24

4.1.3 Double integrals on a rectangle

Finally, we take the limit of the double Riemann sums as the partition is
made finer and finer (so the area ∆A of each sub-rectangle goes to 0) and
define the double integral.
Definition 4.1.3 (Double integral on a rectangle).
Z m X
X n
f (x, y)dA = m→∞
lim f (x∗ij , yij∗ ) ∆A
R n→∞ i=1 j=1

provided the limit exists. Notice that the limit on the right can also be
expressed as
m X
X n N
X
lim
m→∞
f (x∗ij , yij∗ ) ∆A = lim f (x∗k , yk∗ ) ∆A
∆A→0
n→∞ i=1 j=1 k=1

by counting the subrectangles of each partition is some order, since ∆x → 0

as the partitions become more fine.
In the case where the limit exists, we say that f (x, y) is integrable over
R.
RR
Notation 4.1.4. R f (x, y)dA is sometimes uses to denote R the double in-
tegral of f over R. We willRR use the convention of using R f (x, y)dA for a
double integral, and reserve for iterated integrals (which is covered in the
next section).
We might think of the term dA as representing the area of an infinitesimal
rectangle with sides dx and dy.
Example 4.1.3. Using the double Riemann sum approximation from Ex-
ample 4.1.1, we know that
Z p
42 ≈ 16 − x2 − y 2 dA
R

where R : [0, 2] × [0, 2].

CHAPTER 4. HIGHER INTEGRATION 8

4.1.4 Double integrals on a closed and bounded region

We now want to extend the definition of double integrals to any closed and
bounded region, not just rectangles. This can be done by partitioning in-
creasingly larger portions of the region.

• Let R be a closed and bounded region of the plane R2 , with a function

f (x, y) defined on R.
(
a≤x≤b
• Let R : be the smallest rectangle containing R
c≤y≤d
(i.e. a is the minimum of all x-coordinates of points in R, etc.).
• Partition R into subrectangles Rij

, as before. Denote the area of each
subrectangle by ∆A (which is equal to ∆x∆y).
• Now consider only those subrectangles which are contained in R, and
let N be the number of them. We can then order those subrectangles
any way, labeling them now by Rk for k = 1, . . . , N .
• Choose sample points (x∗k , yk∗ ) in each subrectangle Rk .
• A double Riemann sum for this partition is
N
X
f (x∗k , yk∗ )∆A
k=1

Definition 4.1.5 (Double integral on a closed and bounded region).

If the limit of the double Riemann sums exists, as we partition into smaller
and smaller pieces, then f (x, y) is integrable, and we define:
Z N
X
f (x, y)dA = lim f (x∗k , yk∗ )∆A
R ∆A→0
k=1

the double integral of f over R.

Fact 4.1.6. If f is continuous on a closed and bounded region R of R2 , the
f is integrable over R.
More generally, if the value of f on R is bounded, and f is continuous
except possibly at finitely many smooth curves in R, then f is integrable
over R.
CHAPTER 4. HIGHER INTEGRATION 9

Since the volume approximations given by the double Riemann sums keep
improving as the partitions are made with smaller and smaller subrectangles,
we can now answer the motivating problem of computing volume (accommo-
dating for sign).
Let S be a solid in R3 . If S contains points with both positive and
negative z coordinates, then the xy-plane cuts S into pieces which either
above or below. The signed volume V olume± (S) of S is the sum, counting
volumes of pieces above the xy-plane as positive, and volumes of pieces below
as negative.
Fact 4.1.7 (Signed volume of a solid bounded by a graph).
Let f (x, y) be a continuous function defined on a closed and bounded
region R. Then z = f (x, y) for (x, y) in R is a surface. Let S be the solid
bounded by that surface and R in the xy-plane, which consists of all the
points (x, y, z), where (x, y) is in R, and z is between 0 and f (x, y). Then
Z
±
V olume (S) = f (x, y)dA
R

the signed volume of S is equal to the double integral of f over R.

Similarly, if the function gives density, interpreting double Riemann sums
as mass estimations, we get:
Fact 4.1.8 (Mass from a density function).
If R is a closed and bounded region of the xy-plane representing a two-
dimensionalZobject, with density function d(x, y), then the mass of that object
is equal to f (x, y)dA.
R
Example 4.1.4. Compute
Z p
4 − x2 − y 2 dA
R

where R : x2 + y 2 ≤ 4.
p
Solution. We see that R is thepentire domain of f (x, y) = 4 − x2 − y 2 ,
and furthermore the graph z = 4 − x2 − y 2 (over that domain) is exactly
the
p upper hemisphere of a sphere of radius 2. So the volume under z =
4 − x2 − y 2 and above R is half the volume of sphere of radius 2. Using
the geometric formula 43 πr3 for the volume of a sphere of radius r, we can
conclude that: Z p
1 4 16π
4 − x2 − y 2 dA = · π23 =
R 2 3 3
CHAPTER 4. HIGHER INTEGRATION 10

Example 4.1.5. Compute

Z p
9x2 + 9y 2 dA
R

2 2
x + y ≤ 4

where R : x ≥ 0

y≥0


Solution. We can see that portion of the surface

p
z = 9x2 + 9y 2

lying above R is one quarter of an upside-down cone of height 6 and with a

top of radius 2.
Since the volume of a cone of height h and radius r has the geometric
formula 13 πr2 h, the volume underneath an upside-down cone of the same
dimensions has volume πr2 h − 13 πr2 h = 32 πr2 h. Therefore
Z p
2
9x2 + 9y 2 dA = π(22 )(6) = 16π
R 3

4.1.5 Basic properties of double integrals

Fact 4.1.9. Z
Area(R) = 1 dA
R

This follows from the volume of a cylinder being equal to its height times the
area of a its base.
Fact 4.1.10. The average value of an integrable function f (x, y) on a closed
and bounded region R is:
Z
¯ 1
fR = f (x, y) dA
Area(R)
R

Notice that, if f (x, y) ≥ 0 on R then the average value of f on R is the

height of the cylinder with base R, whose volume is equal to the volume of
the solid with base R under the graph of f .
CHAPTER 4. HIGHER INTEGRATION 11

Below we list some simple properties of double integrals. They are very
similar to the properties of ordinary integrals. Let f (x, y) and g(x, y) be
integrable functions on R.

1. Z
For any constants α and β, Z Z
(αf (x, y) + βg(x, y)) dA = α f (x, y) dA + β g(x, y)dA.
R R R
Z Z
2. If f (x, y) ≥ g(x, y) for all (x, y) in R, then f (x, y) dA ≥ g(x, y) dA
R R

3. If the region R is a union of two regions R1 and R2 , whose interiors do

not intersect, then
Z Z Z
f (x, y) dA = f (x, y) dA + f (x, y) dA
R R1 R2

4. If m ≤ f (x, y) ≤ M for all (x, y) in R then

Z
m · Area(R) ≤ f (x, y) dA ≤ M · Area(R)
R

Example 4.1.6. Let R : x2 + y 2 ≤ 1 be the closed unit disc. Using the basic
properties of double integrals, we can show that
Z
4x3 − xy 2 + 3yx2 − 2y 3 dA = 0
R

Solution. Let f (x, y) = 4x3 − xy 2 + 3yx2 − 2y 3 and notice that

f (−x, −y) = −f (x, y). Now split R across the line y = −x into two subre-
gions: (
x2 + y 2 ≤ 1
R1 :
y ≤ −x
(
x2 + y 2 ≤ 1
R2 :
y ≥ −x
CHAPTER 4. HIGHER INTEGRATION 12
Z Z
So f (x, y)dA = − f (x, y)dA and we can conclude that
R1 R2
Z Z Z
f (x, y)dA = f (x, y)dA + f (x, y)dA = 0
R R1 R2
.
CHAPTER 4. HIGHER INTEGRATION 13

4.1.6 Exercises
1. Given the following table of values for f (x, y)

x\y 2 2.5 3 3.5 4

1 8 7 8 9 10
1.25 5 3 4 5 6
1.5 3 3 3 4 5
1.75 1 2 -2 -3 -4
2 3 4 0 -2 -5
Z
Let R : [1, 2] × [2, 4]. Estimate the value of f (x, y)dA by computing:
R

(a) The double Riemann sum for the partition with ∆x = ∆y = 0.5,
with sample points taken at the lower-left corner of each subrect-
angle.
(b) The double Riemann sum for the partition with ∆x = 0.25; ∆y =
0.5, with sample points taken at the lower-left corner of each sub-
rectangle.
(c) The upper double Riemann sum for the partition with ∆x = ∆y =
0.5.
(d) The lower double Riemann sum for the partition with ∆x = ∆y =
0.5.

2. Given the following contour diagram for f (x, y)

CHAPTER 4. HIGHER INTEGRATION 14

2 1 1
10 12 4 6

24 6
28
2
10 8
6
8
1
4 4

22
20
2

18
16
14

8
6

6
0

12
10

2
−1
y

4
−2

4
24 26 28 2
22
20

8
18
16

6
6
14
12

14 12 0
10

1
−3 4
8
6
8
−4 16
−2 −1 0 1 2 3 4
x
Z
Let R : [−2, 0] × [−3, 1]. Estimate the value of f (x, y)dA by com-
R
puting:

(a) The double Riemann sum with ∆x = ∆y = 1 with sample points

taken at the upper-right corner of each subrectangle.
(b) The upper double Riemann sum for the partition with ∆x = ∆y =
1.
(c) The lower double Riemann sum for the partition with ∆x = ∆y =
1.

2 2
x + y ≤ 4

3. Let R : x ≥ y

x≥0

Z p 
Compute 16 − 4x − 4y dA using geometric formulas.
2 2
R
(
x2 + y 2 ≤ 9
4. Let R :
x≤y
Z  p 
Compute −3 + 9 − x2 − y 2 dA using geometric formulas.
R
(
1 ≤ x2 + y 2 ≤ 3
5. Let R :
y≤x
CHAPTER 4. HIGHER INTEGRATION 15

R  p 
Compute R
−1 + x2 + y 2 dA using geometric formulas.
2
Z : x ≤ y ≤ 4. If f (x, y) and g(x, y) are both integrable on R
6. Let R
and |f (x, y) − g(x, y)|dA = 8, what is the average difference of the
R
values f (x, y) and g(x, y) over all (x, y) in R?
CHAPTER 4. HIGHER INTEGRATION 16

Answers
Z
1. (a) f (x, y)dA ≈ 8(0.25) + 3(0.25) + 7(0.25) + 3(0.25) + 8(0.25) +
R
3(0.25) + 9(0.25) + 4(0.25) = 11.25
Z
(b) f (x, y)dA ≈ 8(0.125)+5(0.124)+3(0.125)+1(0.125)+7(0.25)+
R
3(0.125) + 3(0.125) + 2(0.125) + 8(0.125) + 4(0.125) + 3(0.125) +
(−2)(0.125) + 9(0.125) + 5(0.125) + 4(0.125) + (−3)(0.125) = 8.37
Z
(c) f (x, y)dA ≈ 8(0.25) + 4(0.25) + 8(0.25) + 4(0.25) + 9(0.25) +
R
4(0.25) + 10(0.25) + 5(0.25) = 13
Z
(d) f (x, y)dA ≈ 3(0.25)+1(0.25)+3(0.25)+(−2)(0.25)+3(0.25)+
R
(−3)(0.25) + 4(0.25) + (−5)(0.25) = 1
R
2. (a) ds R f (x, y)dA ≈ 10(1)+5(1)+9(1)+4(1)+10(1)+5(1)+13(1)+
8(1) = 64
R
(b) ds R f (x, y)dA ≈ 20(1) + 13(1) + 17(1) + 10(1) + 17(1) + 10(1) +
20(1) + 13(1) = 120
R
(c) ds R f (x, y)dA ≈ 10(1) + 5(1) + 9(1) + 4(1) + 9(1) + 4(1) + 10(1) +
5(1) = 56
R
ds R f (x, y)dA ≈ 20(1) + 13(1) + 17(1) + 10(1) + 17(1) + 10(1) + 20(1) +
13(1) = 120
Z p 
3. 16 − 4x2 − 4y 2 dA = 4π
R
Z  p  9π
4. −3 +
9 − x2 − y 2 dA =
R 2
Z  p   √ 
5. −1 + x2 + y 2 dA = 2 3 − 3 π
R

6. The average value of |f (x, y) − g(x, y)| on R is

Z
1 3
|f (x, y) − g(x, y)| dA =
Area(R) R 4