11

MULTIPLE INTEGRALS

11.1 Introduction
Volume I discussed integrals dx, first for functions defined and bounded on finite
intervals, and later for unbounded functions and infinite intervals. Chapter 10 of Volume
II generalized the concept by introducing line integrals. This chapter extends the concept in
yet another direction. The one-dimensional interval [a, is replaced by a two-dimensional
set Q, called the region of integration. First we consider rectangular regions; later we
consider more general regions with curvilinear boundaries. The integrand is a scalar field f
defined and bounded on Q. The resulting integral is called a double integral and is denoted
by the symbol

or by

As in the one-dimensional case, the symbols dx and dy play no role in the definition of the
double integral; however, they are useful in computations and transformations of integrals.
The program in this chapter consists of several stages. First we discuss the definition of
the double integral. The approach here is analogous to the one-dimensional case treated in
Volume I. The integral is defined first for step functions and then for more general
functions. As in the one-dimensional case, the definition does not provide a useful
procedure for actual computation of integrals. We shall find that most double integrals
occurring in practice can be computed by repeated one-dimensional integration. We shall
also find a connection between double integrals and line integrals. Applications of double
integrals to problems involving area, volume, mass, center of mass, and related concepts
are also given. Finally, we indicate how the concepts can be extended to n-space.

11.2 Partitions of rectangles. Step functions

Let Q be a rectangle, the Cartesian product of two closed intervals [a, b] and [c, d],

Q= = y

An example is shown in Figure 11 .I. Consider two partitions and of [a, b] and [c, d],
353
354 Multiple integrals

Y Y

--

I I
b a

FIGURE 11.1 A rectangle Q, the Cartesian A partition of a rectangle Q.

product of two intervals.

respectively, say

and =

where = a, = = c, = d. The Cartesian product x is said to be a

partition of Since decomposes b] into subintervals and decomposes [c, d] into
m subintervals, the partition P = x decomposes into mn subrectangles. Figure
11.2 illustrates an example of a partition of Q into 30 subrectangles. A partition P’ of Q
is said to be finer than P if P P’, that is, if every point in P is also in P’.
The Cartesian product of two open subintervals of and is a subrectangle with its
edges missing. This is called an open subrectangle of P or of Q.

DEFINITION OF STEP FUNCTION. A function dejined on a rectangle is said to be a step

function if a partition P of exists such that f is constant on each of the open subrectangles

The graph of an example is shown in Figure 11.3. Most of the graph consists of horizontal
rectangular patches. A step function also has well-defined values at each of the boundary

FIGURE 1I The graph of a step function defined on a rectangle Q.

 The double integral of a step function

points of the subrectangles, but the actual values at these points are not relevant to integra-
tion theory.
Iff and g are two step functions defined on a given rectangle Q, then the linear combina-
tion + is also a step function. In fact, if P and P’ are partitions of Q such that is
constant on the open subrectangles of P and g is constant on the open subrectangles of
then + is constant on the open subrectangles of the union P U (which we may
call a common refinement of P and P’). Thus, the set of step functions defined on Q forms
a linear space.

11.3 The double integral of a step function

Let P = x be a partition of a rectangle Q into mn subrectangles and letfbe a
function that is constant on the open subrectangles of Q. Let the subrectangle determined
by and be denoted by and let denote the constant value that
takes at the interior points of Iffis positive, the volume of the rectangular box with
base and altitude is the product

For any step positive or not, the sum of all these products is defined to be the
double integral off over Q. Thus, we have the following definition.

DEFINITION OF THE DOUBLE INTEGRAL OF A STEP FUNCTION. Let be a step function which
takes the constant value on the open subrectangle x rectangle Q.
The double integral off over Q is dejined by the formula

(11.1) jj
Q

As in the one-dimensional case, the value of the integral does not change if the partition
P is replaced by any finer partition P’. Thus, the value of the integral is independent of the
choice of P so long as is constant on the open subrectangles of Q.
For brevity, we sometimes write instead of and instead of
and the sum in (11.1) becomes

To remind ourselves how this sum is formed, we write the symbol for the integral as

This symbol is merely an alternative notation for

Note that if is constant on the interior of say y) = k when a < x < b and
c < y d, we have

(11.2) a)(d
Q
356 Multiple integrals

regardless of the values off on the edges of Q. Since we have

b a= dx and d - c =

formula (11.2) can also be written as

(11.3) = = [ .

The integrals which appear on the right are one-dimensional integrals, and the formula
is said to provide an evaluation of the double integral by repeated or iterated integration.
In particular, when f is a step function of the type described above, we can write

Summing on i and and using (11. we find that (11.3) holds for step functions.
The following further properties of the double integral of a step function are generaliza-
tions of the corresponding one-dimensional theorems. They may be proved as direct
consequences of the definition in (11.1) or by use of formula (11.3) and the companion
theorems for one-dimensional integrals. In the following theorems the symbols and
denote step functions defined on a rectangle Q. To avoid trivial special cases we assume
that Q is a nondegenerate rectangle; in other words, that Q is not merely a single point
or a line segment.

THEOREM LINEARITY PROPERT Y. For every real and we have

jj + = jj Y) + jj Y)
Q Q Q

THEOREM 11.2. ADDITIVE PROPERTY. If Q is subdivided into two rectangles and

then

j j = j j +jj .
Q

11.3 COMPARISON THEOREM. Q,

jj jj .
Q Q

particular, y) 0 for every (x, y) in Q, then

The proofs of these theorems are left as exercises.

 Upper and lower double integrals 357

11.4 The definition of the double integral of a function defined and bounded on a rectangle
Let be a function that is and bounded on a rectangle specifically, suppose
that

Then may be surrounded from above and from below by two constant step functions
and where = and y) = M for all (x, in Q. Now consider any two
step functions s and defined on Q, such that

(11.4) for every point (x, in .

DEFINITION OF THE INTEGRAL OF A BOUNDED FUNCTION OVER A RECTANGLE. If there is one

and only one number Z such that

(11.5) t
Q Q

for every pair of step functions satisfying the inequalities in (I this number I is called
the double integral off over and is denoted by the symbol

f or i s
Q

When such an I exists the function is said to be integrable on

11.5 Upper and lower double integrals

The definition of the double integral is entirely analogous to the one-dimensional case.
Upper and lower double integrals can also be defined as was done in the one-dimensional
case.
Assume is bounded on a rectangle and let s and t be step functions satisfying (11.4).
We say that is below and t is above and we write s t. Let denote the set of all
numbers s obtained ass runs through all step functions below and let T be the set of
all t obtained as t runs through all step functions above Both sets S and T
Q
are since is bounded. Also, t ifs so every number in S is
less than every number in T. Therefore S a and T has an infimum, and they
satisfy the inequalities

Q Q

for all s and t satisfying s t . This shows that both numbers sup and inf T satisfy
(11.5). Therefore, is integrable on if and only if sup = inf T, in which case we have

= sup = inf T.
Q
358 Multiple integrals

The number sup S is called the lower integral off and is denoted by The number
inf T is called the upper integral off and is denoted by Thus, we have

The foregoing argument proves the following theorem.

THEOREM 11.4. Every function which is bounded on a rectangle Q has a lower integral
and an upper integral satisfying the inequalities

Q Q

all step functions and t with t . The function is integrable on Q if and only
its upper and lower integrals are equal, in which case we have

= =
Q

Since the foregoing definitions are entirely analogous to the one-dimensional case, it is
not surprising to learn that the linearity property, the additive property, and the comparison
theorem as stated for step functions in Section 11.3, also hold for double integrals in general.
The proofs of these statements are analogous to those in the one-dimensional case and will
be omitted.

11.6 Evaluation of a double integral by repeated one-dimensional integration

In one-dimensional integration theory, the second fundamental theorem of calculus
provides a practical method for calculating integrals. The theorem accomplishes the
same result in the two-dimensional theory; it enables us to evaluate certain double integrals
by means of two successive one-dimensional integrations. The result is an extension of
formula which we have already proved for step functions.

THEOREM 11.5. Let be defined and bounded on a rectangle Q = [a, b] x [c, d] , and
assume that is integrable on Q. For y in [c, d] assume that the one-dimensional
integral y) dx exists, and denote its value by A(y). If the integral A(y) dy exists it
is equal to the double integral In other words, we have the formula
Q

(11.6) =

Proof. Choose any two step functions and satisfying on Q. Integrating

with respect to over the interval [a, b] we have
 Geometric interpretation of the double integral as a volume 359

Since the integral A(y) dy exists, we can integrate both these inequalities with respect to
over [c, d] and use Equation (11.3) to obtain

Therefore A(y) dy is a number which lies between and for all step functions
and approximating from below and from above, respectively. Since is integrable on
Q, the only number with this property is the double integral off over Therefore
A(y) dy = which proves Equation (11.6).

Formula (11.6) is said to provide an evaluation of the double integral by repeated or

iterated integration. The process is described by saying that first we integrate with respect
to from a to (holding fixed), and then we integrate the result with respect to from
c to d. If we interchange the order of integration, we have a similar formula, namely,

(11.7)

which holds if we assume that y) dy exists for each fixed in [a, b] and is integrable
on [a, b].

Theorem 11.5 has a simple geometric interpretation, illustrated in Figure 11.4. If is

nonnegative, the set of points (x, y, in 3-space with (x, y) in Q and 0 z is

The ordinate set of f’over Cross section with area

FIGURE 11.4 The volume of is the integral of the cross-sectional area:

360 Multiple integrals

called the ordinate set off over Q. It consists of those points between the rectangle Q and
the surface z = f (x, y) . (See Figure 11.4(a).) For each in the interval [c, d], the integral

is the area of the cross section of cut by a plane parallel to the xz-plane (the shaded region
in Figure 11.4(b)). Since the cross-sectional area A(y) is integrable on [c, d], Theorem 2.7
of Volume tells us that the integral A(y) is equal to v(S), the volume of S. Thus, for
nonnegative integrands, Theorem 11.5 shows that the volume of the ordinate set off over Q
is equal to the double integral
Equation (11.7) gives way of computing the volume of the ordinate set. This
time we integrate the area of the cross sections cut by planes parallel to the yz-plane.

11.8 Worked examples

In this section we illustrate Theorem 11.5 with two numerical examples.

EXAMPLE If Q = [-1, 1] X evaluate (x sin ye”) dx given that

the integral exists. The region of integration is Figure 11.5.

Solution. Integrating first with respect to x and calling the result A(y), we have

= x sin y ye”) dx = sin y - y e ” = - e y + y / e .

2

FIGURE 11.5 The region of integration for FIGURE 11.6 The region of integration for
Example 1. Example 2.
 Worked examples 361

Applying Theorem 11.5 we find

(x sin ye”) dx dy = = +

= (l/e e) y dy = (l/e

As a check on the calculations we may integrate first with respect to y:

(x sin y ye”) dx dy = [ sin y dy]

cos y dx

+ x) dx = (l/e .

EXAMPLE 2. If Q [- 1, x evaluate the double integral dx ,

given that it exists.

Solution. If we integrate first with respect toy and call the result H(x), we have H(x) =
dy . The region of integration is the rectangle shown in Figure 11.6. The
parabola y = is also shown because of the presence of in the integrand. Above
this parabola we have y > and below it we have y < x This suggests that we split the
integral for H(x) as follows:

We remember that is treated as a constant in each of these integrals. In the first integral
we make the change of variable = y and in the second we put = y This
gives us
dy dt + dt = +

Applying Theorem 11.5 we find

dx dy = + dx = (2
0

+ + 6 + .
3 x(2 0 3 2

The same result may be obtained by integrating first with respect to x, but the calculations
are more complicated.
 Multiple integrals

Evaluate the double integrals in Exercises 1 through 8 by repeated integration, given that each
integral exists.

1. , where Q = [0, x [0,

Q

where Q =

dx dy , where Q = [0,

x y dx dy , where Q =

5. where Q =
i s
Q
where Q =

where Q= x and denotes the greatest integer

Q

dx dy, where Q = x >O.

Q
9. If Q is a rectangle, show that a double integral of the form dx dy is equal to the
product of two one-dimensional integrals. State the about existence that you are
using.
10. Letfbe defined on the rectangle Q = [0, x [0, as follows:

1 -x-y i f
=
otherwise.

Make a sketch of the ordinate set off over Q and compute the volume of this ordinate set by
double integration. (Assume the integral exists.)

11. Solve Exercise 10 when

i f
otherwise.

12. Solve Exercise 10 when Q = [ -1, x [ -1, and

if + 1,
 Integrability of continuous functions 363

13. Let be defined on the rectangle Q = x as follows:

+ i f
=
otherwise.

Indicate, by means of a sketch, the portion of Q in which is and compute the value
of the double integral iven that the integral exists.
14. Let be defined on Q = [0, x [0, as follows:

i f

Prove that the double integral and equals 0.

11.10 Integrability of continuous functions

The small-span theorem (Theorem 9.10) can be used to prove integrability of a function
which is continuous on a rectangle.

THEOREM 11.6. I N T EG R A B IL IT Y OF CO NT IN U OU S FU NC TI O NS . Ifafinction is continuous

on a rectangle = [a, b] x [c, then is integrable on Q. Moreover, the value of the
integral can be obtained by iterated integration,

= =
Proof. Theorem 9.8 shows that is bounded on Q, so has an upper integral and a
lower integral. We shall prove that = Choose > 0. By the small-span
theorem, for this choice of there is a partition P of Q into a finite number (say n) of
rectangles .. . , such that the span every subrectangle is less than Denote
respectively, the absolute maximum and minimum values off in
Then we have

Now let and be two step functions defined on the interior of

each as follows:

i f

At the boundary points we define s(x) = m and = M, where m and Mare, respectively,
the absolute minimum and maximum values off on Then we have for all
in Also, we have

Q .
364 Multiple integrals

where is the area of rectangle . The difference of these two integrals is

k=l
i
Q Q

where a(Q) is the area of Since we obtain the inequality

Q

Letting 0 we see that = so is integrable on

Next we prove that the double integral is equal to the first iterated integral in (11.8).
For each fixed in [c, the one-dimensional integral f (x, exists since the integrand
is continuous on Let A(y) = dx . We shall prove that A is continuous on
[c, d]. Ify are any two points in [c, d] we have

from which we find.

where is a point in [a, where (x, attains its maximum. This inequality
shows that A(y) as so A is continuous at Therefore the integral
A(y) dy exists and, by Theorem 11.5, it is equal to A similar argument works when
the iteration is taken in the reverse order.

11.11 Integrability of bounded functions with discontinuities

Let be defined and bounded on a rectangle In Theorem 11.6 we proved that the
double integral off over exists if f is continuous everywhere on In this section we
prove that the integral also exists if f has discontinuities in provided the set of
continuities is not too large. To measure the size of the set of discontinuities we introduce
the following concept.

DEFINITION OF A BOUNDED SET OF CONTENT ZERO. Let A be a bounded subset of the plane.
The set A is said to have content zero every > 0 there is set of rectangles whose
union contains A and the sum of whose areas does not exceed

In other words, a bounded plane set of content zero can be enclosed in a union of rec-
tangles whose total area is arbitrarily small.
The following statements about bounded sets of content zero are easy consequences of
this definition. Proofs are left as exercises for the reader.
(a) Any finite set of points in the plane has content zero.
(b) The union of a finite number of bounded sets of content zero is also of content zero.
(c) Every subset of a set of content zero has content zero.
(d) Every line segment has content zero.
 Double integrals extended over more general regions 365

THEOREM 11.7. Let be and bounded on a rectangle Q = [a, b] x [c, d]. If the
set of discontinuities off in Q is a set of content zero then the double integral exists.
Q

Proof. Let M > 0 be such that M on Q. Let D denote the set of discontinuities
off in Q. Given > 0, let P be a partition of Q such that the sum of the areas of all the
subrectangles of P which contain points of D is less than (This is possible since D has
content zero.) On these subrectangles define step functions and t as follows:

s(x) = -M, t(x) = M.

On the remaining subrectangles of P define and t as was done in the proof of Theorem 11.6.
Then we have t throughout Q. By arguing as in the proof of Theorem 11.6 we
obtain the inequality

(11.9) +
Q Q

The first term, comes from estimating the integral of t over the subrectangles
containing only points of continuity off; the second term, comes from estimating
the integral of t over the subrectangles which contain points of D. From (11.9) we
obtain the inequality
+

Letting 0 we find 0 i(f) Since is arbitrary this implies =

so f is integrable on Q.

11.12 Double integrals extended over more general regions

Up to this point the double integral has been defined only for rectangular regions of
integration. However, it is not difficult to extend the concept to more general regions.
Let S be a bounded region, and enclose in a rectangle Q. Let be defined and bounded
on S. Define a new function on Q as follows :

(11.10)
if -S.

In other words, extend the definition off to the whole rectangle Q by making the function
values equal to 0 outside S. Now ask whether or not the extended function is integrable
on Q. If so, we say that f is integrable on S and that, by dejinition,

First we consider sets of points S in the xy-plane described as follows:

and y
366 Multiple integrals

where and are functions continuous on a closed interval [a, and satisfying
An example of such a region, which we call a region of Type I, is shown in Figure 11.7.
In a region of Type I, for each point t in [a, the vertical line = t intersects S in a line
segment joining the curve = to = Such a region is bounded because
and are continuous and hence bounded on [a, h].
Another type of region T (Type II) can be described as follows:

where and are continuous on an interval [c, d] with An example is shown

in Figure 11.8. In this case horizontal lines intersect T in line segments. Regions of Type
II are also bounded. All the regions we shall consider are either of one of these two types or
can be split into a finite number of pieces, each of which is of one of these two types.

FIGURE 11.7 A region of Type I FIGURE 11.8 A region T of Type 11.

Let be defined and bounded on a region S of Type I. Enclose in a rectangle and

definefon Q as indicated in Equation (11.10). The discontinuities Q will consist of
the discontinuities off in S plus those points on the boundary of at which is
The boundary of S consists of the graph of the graph of and two vertical line
segments joining these graphs. Each of the line segments has content zero. The next
theorem shows that each of the graphs also has content zero.

THEOREM 11.8. Let be a real-valued function that is continuous on an interval [a, b].
Then the graph of has content zero.

Proof. Let A denote the graph of that is, let

 Double integrals extended over more general regions 367

Choose any 0. We apply the small-span theorem (Theorem 3.13 of Volume I) to

obtain a partition P of [a, b] into a finite number of subintervals such that the span of in
every subinterval of P is less than a). Therefore, above each subinterval of P the
graph of lies inside a rectangle of height a). Hence the entire of graph lies
within a finite union of rectangles, the sum of whose areas is (An example is shown in
Figure 11.9.) This proves that the graph of has content zero.

FIGURE 11.9 Proof that the graph of a continuous function has content zero.

The next theorem shows that the double integral exists continuous on int S, the
interior of S. This is the set

int S = a < .

THEOREM 11.9. Let S be a region of Type I, between the graphs of and Assume that
is and bounded on S and that is continuous on the interior of S. Then the double
integral f exists and can be evaluated by repeated one-dimensional integration,

(11.11)

Proof. Let Q = [a, b] x [c, d] be a rectangle which contains S, and let be defined by
(11.10). The only possible points of discontinuity the boundary points of S. Since
the boundary of S has content integrable on Q. For each fixed x in (a, the
dimensional integral dy exists since the integrand has at most two discontinuities
on [c, d]. Applying the version of Theorem 11.5 given by Equation (11.7) we have

(11.12)
368 Multiple integrals

= 0 for the remaining values of

in [c, d]. Therefore

=
so Equation (11.12) implies (11.11).

There is, of course, an analogous theorem for a region T of Type II. Iffis defined and
bounded on T and continuous on the interior of then is integrable on T and the formula
for repeated integration becomes

= dx] dy .

Some regions are of both Type I and Type II. (Regions bounded by circles and ellipses
are examples.) In this case the order of integration is immaterial and we may write

[
In some cases one of these integrals may be much easier to compute than the other; it is
usually worthwhile to examine both before attempting the actual evaluation of a double
integral.

Let be a region of Type I given by

s = {(x, a x

Applying Theorem 11.9 with (x, = 1 for all (x, in S we obtain

jj =
S

By Theorem 2.1 of Volume I, the integral on the right is equal to the area of the region S.
Thus, double integrals can be used to compute areas.
If is nonnegative, the set of points (x, z) in 3-space such that (x, S and
0 (x, is called the ordinate set off over S. An example is shown in Figure 11.10.
Iff is nonnegative and continuous on the integral

represents the area of a cross-section of the ordinate set cut by a plane parallel to the
yz-plane (the shaded region in Figure 10). Formula (11.11) of Theorem 11.9 shows that
the double integral off over is equal to the integral of the cross-sectional area. Hence
 Worked examples 369

section

FIGURE 11.10 The integral is e area of a cross section of the ordinate

set. The iterated integral is the volume of the ordinate set.

the double integral equal to the volume of the ordinate set (See Theorem
2.7 of Volume I,
More generally, if and are both continuous on with , then the double integral
is equal to the volume of the solid lying between the graphs of the functions
and g. Similar remarks apply, of course, to regions of Type II.

11.14 Worked examples

EXAMPLE Compute the volume of the solid enclosed by the ellipsoid

Solution. The solid in question lies between the graphs of two functions and where

= =

Here (x, y) where is the elliptical region given by

= + 1 .

Applying Theorem 11.9 and using the symmetry we find that the volume of the ellipsoidal
solid is given by

V 2 g dy]
S S
370 Multiple integrals

Let A = Then the inner integral is

A dy .

Using the change of variable = Ab sin dy = Ab cos t dt , we find that the last integral
is equal to

Therefore

In the special case a = b = c the solid is a sphere of radius a and the volume is

EXAMPLE 2. The double integral of a positive function f, y) dx reduces to the

repeated integral S

Determine the region and interchange the order of integration.

Solution. For each fixed x between 0 and 1, the integration with respect to y is over
the interval from to x. This means that the region is of Type I and lies between the two
curves y = andy = x. The region is the set of points between these two curves and
above the interval [0, (See Figure 11 1.) Since is also of Type II we may interchange

FIGURE 11.11 Example 2. 11.12 Example 3.

 Exercises 371

the order of integration to obtain

EXAMPLE 3. A double integral of a positive function y) dx reduces to the

repeated integral : s
- -
3
.

Determine the region S and interchange the order of integration.

Solution. For each fixed y between 0 and 3, the integration with respect to x is over
the interval from to Therefore region S is of Type II and lies between
the two curves x = and x = This region, shown in Figure 11.12, is a
sector of a circle. When the order of integration is reversed the region must be split into
two regions of Type I; the result is the sum of two integrals:

+ dx.

11.15 Exercises
In Exercises 1 through 5, make a sketch of the region of integration and evaluate the double
integral.

1. x cos (x + y) dx , where is the triangular region whose vertices are (0, 0), 0), (x, x).
s

2. (1 + x) siny dx where is the trapezoid with vertices (0, 0), (1, 0), (0, 1).
S

3. dx dy, where = {(x, y) I .

S

4. where S is the bounded portion of the first quadrant lying between the two

xy = 1 and xy = 2 and the two straight lines y = and y = 4x.

5. (x” dx dy , where S is bounded by the curve y = sin x and the interval [0,
S

6. A pyramid is bounded by the three coordinate planes and the plane x + + = 6. Make
a sketch of the solid and compute its volume by double integration.
7. A solid is bounded by the surface z = x the xy-plane, and the planes x = 1 and x = 3.
Make a sketch of the solid and compute its volume by double integration.
8. Compute, by double integration, the volume of the ordinate set S if:
= + and =
= + y and
=y +2x and S