Chapter 4

Higher integration

4.2 Iterated integrals

4.2.1 Iterated integrals over rectangular regions
(
a≤x≤b
Definition 4.2.1. Let R : , with f (x, y) continuous on R.
c≤y≤d
The expression Z Z d b
f (x, y)dxdy
c a
is called an iterated integral.
This notation is taken to mean
Z dZ b Z d Z b 
f (x, y)dxdy = f (x, y)dx dy
c a c a

Similarly,
Z bZ d Z b Z d 
f (x, y)dydx = f (x, y)dy dx
a c a c

is also an iterated integral.

Let’s examine what such expressions representR in closer detail. Notice
b
that, fixing a value of y ie. y = y0 , means that a f (x, y0 )dx is a regular
definite integral of the single variable function f (x, y0 ), and the value of
Rb Rb
a
f (x, y0 )dx depends on y 0 only. In other words, we can think of a
f (x, y)dx

1
CHAPTER 4. HIGHER INTEGRATION 2
Rb
as a function of y, g(y) = a f (x, y)dx. This function can then be integrated,
Rd RdRb
ie. c g(y)dy. In other words, c a f (x, y)dxdy is just two iterations of
integrals of single-variable functions.
Then, iterated integrals require no special methods to compute, other
than the standard techniques for integrating single-variable functions. Like
partial differentiation, we just need to keep track of what is being treated
as a variable, and what is being treated as constant/fixed. Let’s look at an
example of computing an iterated integral.
R0 R1
Example 4.2.1. Compute −2 0 (x2 y + xy 3 + y 2 − 2x) dxdy
Solution. Since
Z 0Z 1 Z 0 Z 1 
2 3 2 2 3 2



x y + xy + y − 2x dxdy = x y + xy + y − 2x dx dy
−2 0 −2 0
R1
we need to first compute 0
(x2 y + xy 3 + y 2 − 2x) dx.
Z 1
x2 y + xy 3 + y 2 − 2x dx


0
 3 1
x y x2 y 3 2 2
= + + xy − x
3 2 0
3
y y
= + + y2 − 1
3 2
So substituting this function of y in, we can complete our computation
of the iterated integral.
Z 0Z 1
x2 y + xy 3 + y 2 − 2x dxdy


−2 0
Z 0 Z 1 
2 3 2


= x y + xy + y − 2x dx
−2 0
Z 0
y y3

2
= + + y − 1 dy
−2 3 2
 2 0
y y4 y3
= + + −y
6 8 3 −2
=2
CHAPTER 4. HIGHER INTEGRATION 3

The relationship between double integrals and iterated integrals

Now that we know how to read and compute iterated integrals, let’s turn our
attention to their graphical interpretation. From this, we will see why they
are relevant to us now (in the context of double integrals).
To make matters simpler, let’s assume for now that f (x, y) ≥ 0 on R.
Now consider the solid S bounded by the rectangle R : [a, b] × [c, d] below
and by the graph z = f (x, y) above.

20

15

10

5
−1
0
−1 0
−0.5 0
0.5 1 1

In this case R : [−1, 1] × [−1, 1] and f (x, y) = −3x3 + x2 + 8y 2 x − 2y 3 + 10.

Note that f (x, y) > 0 on R. So the signed volume of S is the just the regular
volume.
Let’s start by fixing y = yj for some c ≤ yj ≤ d, and try visualize
Rb
a
f (x, yj )dx. We’ve already seen (when interpreting the partial derivatives
of f ) that z = f (x, yj ) is the cross-sectional curve of z = f (x, y) in the plane
Rb
y = yj . So then a f (x, y0 )dx can be seen as the area of the cross-sectional
slice through the solid S by the plane y = yj .
CHAPTER 4. HIGHER INTEGRATION 4

20

15

10

5
−1
0
−1 0
−0.5 0
0.5 1 1

Suppose we partition the y interval [−1, 1] into 8 subintervals with end-

points y0 = −1, y1 = −0.75, . . . , y8 = 1 (so ∆y = 0.25).

Then the volume of the orange solid is

Z 1 
f (x, y2 )dx ∆y
0

where yj = −0.5 and ∆y = 0.25. So we see that

Xn Z 1 
f (x, yj )dx ∆y ≈ V olume(S)
j=1 −1
CHAPTER 4. HIGHER INTEGRATION 5

.
By taking the limit as n → ∞, we can conclude that
Z 1Z 1
f (x, y)dxdy = V olume(S)
−1 −1

. But we already know that the value of the double integral of f over R is
equal to V olume(S). So we’ve justified the following fact.
Fact 4.2.2 (Fubini’s Theorem for(rectangular regions).
a≤x≤b
If f (x, y) is continuous on R : , then:
c≤y≤d
Z Z d Z b Z bZ d
f (x, y)dA = f (x, y)dxdy = f (x, y)dydx
R c a a c

Even more generally, this equality still holds when f (x, y) is any function
that is bounded on R, is continuous except possibly on finitely many smooth
curves in R, and where the iterated integrals on the right exist.
Recall that previously, we could only estimate the value of most double
integrals (with Riemann sums). Now, we don’t need to rely on using ”known”
geometric volume formulas (in fact, we can prove many of those now!). Fu-
bini’s Theorem gives us a practical method for reduce double integrals to
iterated integrals, to which we can apply all of our single-variable integration
techniques.
To understand why the double integral of f over R has the same value as
the iterated integral of f with bounds from the corresponding
Z Z x and y inter- d b
vals, it helps to have some graphical interpretation of f (x, y)dxdy.
c a
Example 4.2.2. Compute the volume of the solid S bounded between the
surface z = −3x3 +x2 +8y 2 x−2y 3 +10, and the 2 by 2 square in the xy-plane
centered at the origin.
Solution. From the previous section, we know that
Z
V olume(S) = −3x3 + x2 + 8y 2 x − 2y 3 + 10dA
R

where R : [−1, 1] × [−1, 1].

Now we know that
CHAPTER 4. HIGHER INTEGRATION 6

Z
V olume(S) = −2x3 + 8xy + y 2 − 3y 3 + 10dA
ZR1 Z 1
−2x3 + 8xy + y 2 − 3y 3 + 10 dxdy


=
−1 −1

and we can compute

Z 1 Z 1
−2x3 + 8xy + y 2 − 3y 3 + 10 dxdy


−1 −1
Z 1  1 Z 1
1 4 2 2 3
= − x + 4x y + xy − 3xy + 10x dxdy = 2y 2 − 6y 3 + 20dy
−1 2 −1 −1
 1
2 6
= y 3 − y 4 + 20y
3 4 −1
4
= + 40
3
124
=
3
Remember that Fubini’s Theorem says that for f continuous on a rect-
angular region R,
Z Z dZ b Z bZ d
f (x, y)dA = f (x, y)dxdy = f (x, y)dydx
R c a a c

in particular, we can switch the order of the iterated integral. It’s important
to note that this is only so straightforward because the region is rectangu-
lar. The main reason we would choose one order versus the other is for
computational convenience.
Example 4.2.3 (Switching the order of an iterated integral over a rectan-
gular region). Z
Compute the double integral y sin(xy) dA, where R = [1, 2] × [0, π].
R
Solution. Here, if we integrate with respect to x first then we get
Z Z πZ 2
y sin(xy) dA = y sin(xy) dx dy
0 1
R
CHAPTER 4. HIGHER INTEGRATION 7

and making the substitution u = xy in the inner integral we obtain

Z Z π 2 Z Z Z "
π
#
2y π 2y
y sin(xy) dx dy = sin u du dy = − cos u dy
0 1 0 y 0 y
Z π Z π Z π
= (− cos(2y) + cos y) dy = cos y dy − cos(2y) dy
0 0 0
π π
1
= sin y − · sin(2y) =0
0 2 0
Now, if we integrate with respect to y first then we get
Z Z 2Z π
y sin(xy) dA = y sin(xy) dy dx
1 0
R

and in order to compute the inner integral

Z π
y sin(xy) dy
0

we
Z have to apply integration by parts. So, the whole process of computing
y sin(xy) dA is going to be much more involved than in the first solution
R
(although with the same result).

4.2.2 Iterated integrals over closed and bounded re-

gions
Consider the integral Z
I= f (x, y) dA
R

over the region R.

CHAPTER 4. HIGHER INTEGRATION 8

If R is not a rectangle then reduction of I to an iterated integral of the

type Z bZ d Z dZ b
f (x, y) dy dx or f (x, y) dx dy
a c c a
is impossible. But if the boundary of R is given by a finite set of smooth
curves then one can reduce I to iterated integrals of special type described
below.

Vertically or horizontally sliceable regions

A vertically sliceable region (or region of Type I) can be represented in the
form
R = {(x, y) | a 6 x 6 b, g1 (x) 6 y 6 g2 (x)}
That is, R is bounded above and below by the curves y = g1 (x) and y = g2 (x)
and on the sides by the vertical lines x = a and x = b.

Here, for every fixed x in the interval [a, b], the values of y change between
g1 (x) and g2 (x) (that is, R is sliced vertically). Hence, the double integral of
a function f (x, y) over R can be reduced to an iterated integral as follows
Fact 4.2.3 (Fubini’s Theorem for vertically sliceable regions).
Z Z b Z g2 (x)
f (x, y) dA = f (x, y) dy dx
a g1 (x)
R

Similarly we can define a horizontally sliceable region (or region of Type

II) which has the form

R = {(x, y) | c 6 y 6 d, h1 (y) 6 x 6 h2 (y)}

In this case, R is bounded above and below by the horizontal lines y = c and
y = d, and on the sides by the curves x = h1 (y) and x = h2 (y). Observe that
the side curves are given as functions of y.
CHAPTER 4. HIGHER INTEGRATION 9

Thus, for every fixed y in the interval [c, d], the values of x change between
h1 (y) and h2 (y) (that is, R is sliced horizontally), and the double integral of
a function f (x, y) over R can be reduced to an iterated integral as follows
Fact 4.2.4 (Fubini’s Theorem for horizontally sliceable regions).
Z Z d Z h2 (y)
f (x, y) dA = f (x, y) dx dy
c h1 (y)
R

Why do we need regions of Type I and Type II? The idea is that every
region can be split into a union of such regions. One can do it by cutting
the given region R into sub-regions using horizontal and vertical lines. This
means we have a method for computing double integrals over any closed
and bounded region. We just need to decompose the region into sliceable
sub-regions, and then compute the corresponding iterated integrals.

For example, the region R shown above can be split into three regions
R1 , R2 , and R3 , where R1 and R3 are regions of Type II and R2 is a region
of Type I. Hence, by the properties of double integrals, the double integral
of a function f (x, y) over R splits into the sum of three double integrals
Z Z Z Z
f (x, y) dA = f (x, y) dA + f (x, y) dA + f (x, y) dA
R R1 R2 R3

and the new double integrals can reduced to iterated ones.

CHAPTER 4. HIGHER INTEGRATION 10
Z
Example 4.2.4. Compute (x + 2y) dA, where R is bounded by the curves
R
y = 2x2 and y = 1 + x2 .
Solution: The region R is shown below

This is a region of Type I and we have to find the bounds in which x

changes. To do this we compute the intersection points of the given parabo-
las. The equation 2x2 = 1 + x2 has two solutions x = 1 and x = −1, hence,
−1 6 x 6 1 and for each x in this interval, y changes from 2x2 (lower
boundary of R) to 1 + x2 (upper boundary of R).
Hence, we have
Z 1" 1+x2
Z Z 1 Z 1+x2 #
xy + y 2


(x + 2y) dA = (x + 2y) dy dx = dx
−1 2x2 −1 2x2
R

Z 1
x(1 + x2 ) + (1 + x2 )2 − (x(2x2 ) + (2x2 )2 dx
 
=
−1
1  1
x2 2 3 x4 3 5
Z 
2 3 4
= (1 + x + 2x − x − 3x ) dx = x + + x − − x
−1 2 3 4 5 −1
   
1 2 1 3 1 2 1 3 4 6 32
= 1+ + − − − −1 + − − + =2+ − =
2 3 4 5 2 3 4 5 3 5 15
Z
Example 4.2.5. Compute xy dA, where R is bounded by the curves
R
y = x − 1 and y 2 = 2x + 6.
Solution: Observe that the equation y 2 = 2x + 6 can be rewritten as
1
x = y 2 − 3 which describes a parabola whose axis coincides with the x-axis.
2
Similarly, we can rewrite y = x − 1 as x = y + 1. Hence, the region R can be
considered as a Type II region of the following shape
CHAPTER 4. HIGHER INTEGRATION 11

1
Here, the left bound is the parabola x = y 2 − 3 and the right bound is
2
the line x = y + 1.
Next, we find the bounds in which y changes. Compute the intersection
1
points of the given curves: the equation y 2 − 3 = y + 1 has two solutions
2
x = 4 and x = −2. Hence, −2 6 y 6 4 and for each y in this interval, x
changes from 21 y 2 − 3 (left boundary of R) to y + 1 (right boundary of R).
Thus, we have
Z Z 4 Z y+1 Z 4 " y+1
#
y 2
xy dA = xy dx dy = x dy
−2 12 y 2 −3 −2 2 1 2
y −3
R 2

Z 4  2 !
y 1
= (y + 1)2 − (y 2 − 3) dy
−2 2 2
4
1 4
Z  5   6 
y 3 2 1 y 4 2 3 2
= − + 4y + 2y − 8y dy = − + y + y − 4y
2 −2 4 2 24 3 −2

= 36
Sometimes the proposed order of integration in an iterated integral is not
convenient. In such cases switching the order of integration may help.
Z 1Z 1
Example 4.2.6. Compute sin(y 2 ) dy dx.
Z 0 x
Solution: The integral sin(y 2 ) dy cannot be expressed as a finite sum
of
Z 1elementary
Z 1 functions. Hence, it is impossible to compute the integral
sin(y 2 ) dy dx with the proposed order of integration: at first by y and
0 x
then by x. Consider the region of integration R: this is a Type I region

R = {(x, y) | 0 6 x 6 1, x 6 y 6 1}
CHAPTER 4. HIGHER INTEGRATION 12

Observe that R can be also interpreted as a Type II region

R = {(x, y) | 0 6 y 6 1, 0 6 x 6 y}

so, we have
Z 1Z 1 Z Z 1 Z y
2 2
sin(y ) dy dx = sin(y ) dA = sin(y 2 ) dx dy
0 x 0 0
R

and the latter integral is not hard to compute. We have

Z 1 Z y Z 1 1
2 1 1
sin(y ) dx dy = 2
y sin(y ) dy = − cos(y 2 ) = (1 − cos 1)
0 0 0 2 0 2
CHAPTER 4. HIGHER INTEGRATION 13

4.2.3 Exercises
1. Compute the iterated integral:
Z 1 Z 2
(a) (x + 2y + xy + 1) dxdy
0 0
Z 1 Z 2
(b) (x + 2y + xy + 1) dydx
0 0
Z 3 Z y
(c) (xy)dxdy
1 0
√
Z 1 Z x
y
(d) dydx
0 0 ex2
π
Z
2
Z cos θ
(e) r (cos θ − r) drdθ
0 0

2. Change the order of the iterated integral and evaluate:

Z √π Z √π
2 2
cos y 2 dydx


(a)
0 x
Z 1 Z 1
4y
(b) dxdy
0 y x2 + y2
Z 2 Z √
3 y
2 2
(c) ex dxdy
y
0 2

3. Compute the double integral:

(
0≤x≤2
Z
(a) (12 − 6x − 4y) dA, where R :
R 0≤y≤1
Z
x2 − y 2 dA, where R is the rectangle with corners (±1, ±).


(b)
R
Z
√
(c) xy 2 dA, where R : y 3 ≤ x ≤ 3 y
R
Z
x2 y − x dA, where R is the lower half of the closed disk of


(d)
R
radius 3.
CHAPTER 4. HIGHER INTEGRATION 14
Z
y−1
(e) ex dA , where R is bounded by x = ln y and x = e−1
.
R

4. Find the volume of the solid:

√
(a) under the surface z = 1 − x2 and above the triangle in the xy-
plane formed by y = x, y = −x, and x = 1.
2
(b) under the surface z = ex and above the triangle with vertices
(0, 0, 0), (1, 1, 0), (1, 3, 0))

5. Suppose an two-dimensional object is represented by the region bounded

by x = 0, x = y 2 , and y = 2 in the xy-plane (measured in meters). If
kg
the density of the object at (x, y) is given by d(x, y) = yex m 2 , then

what is the object’s mass?

CHAPTER 4. HIGHER INTEGRATION 15

4.2.4 Answers
1. (a) 7
(b) 8
(c) 10
1 1


(d) 4
1− e
1
(e) 9
1
2. (a) 2
(b) 2 ln(2)
(c) e − 2

3. (a) 8
(b) 0
8
(c) 99
−162
(d) 5
1
(e) (1 − e)e e−1 + e
2
4. (a) 3
.
(b) e − 1.
1
5. Mass = 2
(e4 − 3) kg