15.

8 Substitutions in Multiple Integrals 905

Substitutions in Multiple Integrals

15.8
The goal of this section is to introduce you to the ideas involved in coordinate transforma-
tions. You will see how to evaluate multiple integrals by substitution in order to replace
v complicated integrals by ones that are easier to evaluate. Substitutions accomplish this by
simplifying the integrand, the limits of integration, or both. A thorough discussion of mul-
(u, v) tivariable transformations and substitutions, and the Jacobian, is best left to a more ad-
G vanced course following a study of linear algebra.

Substitutions in Double Integrals

u
0 The polar coordinate substitution of Section 15.4 is a special case of a more general sub-
stitution method for double integrals, a method that pictures changes in variables as trans-
Cartesian uv-plane formations of regions.
Suppose that a region G in the uy-plane is transformed one-to-one into the region R in
x 5 g(u, v)
y 5 h(u, v) the xy-plane by equations of the form
x = gsu, yd, y = hsu, yd,
y
as suggested in Figure 15.53. We call R the image of G under the transformation, and G
the preimage of R. Any function ƒ(x, y) defined on R can be thought of as a function
(x, y) ƒ(g(u, y), h(u, y)) defined on G as well. How is the integral of ƒ(x, y) over R related to the
R
x integral of ƒ(g(u, y), h(u, y)) over G?
0
The answer is: If g, h, and ƒ have continuous partial derivatives and J(u, y) (to be dis-
cussed in a moment) is zero only at isolated points, if at all, then

6 6
Cartesian xy-plane ƒsx, yd dx dy = ƒsgsu, yd, hsu, ydd ƒ Jsu, yd ƒ du dy. (1)
FIGURE 15.53 The equations R G
x = gsu, yd and y = hsu, yd allow us to The factor J(u, y), whose absolute value appears in Equation (1), is the Jacobian of
change an integral over a region R in the the coordinate transformation, named after German mathematician Carl Jacobi. It meas-
xy-plane into an integral over a region G ures how much the transformation is expanding or contracting the area around a point in G
in the uy-plane by using Equation (1). as G is transformed into R.

DEFINITION The Jacobian determinant or Jacobian of the coordinate

transformation x = gsu, yd, y = hsu, yd is

Jsu, yd = 4 4 = 0x
0x 0x
0u 0y 0y 0y 0x
- . (2)
0y 0y 0u 0y 0u 0y
0u 0y

The Jacobian can also be denoted by

0sx, yd
Jsu, yd =
0su, yd
HISTORICAL BIOGRAPHY to help us remember how the determinant in Equation (2) is constructed from the partial
Carl Gustav Jacob Jacobi derivatives of x and y. The derivation of Equation (1) is intricate and properly belongs to a
(1804–1851) course in advanced calculus. We do not give the derivation here.

y = r sin u, and use Equation (1) to write the Cartesian integral 4R ƒ(x, y) dx dy as a
EXAMPLE 1 Find the Jacobian for the polar coordinate transformation x = r cos u,

polar integral.
906 Chapter 15: Multiple Integrals

u Solution Figure 15.54 shows how the equations x = r cos u, y = r sin u transform the
rectangle G: 0 … r … 1, 0 … u … p>2, into the quarter circle R bounded by x 2 + y 2 = 1
p in the first quadrant of the xy-plane.
2 For polar coordinates, we have r and u in place of u and y. With x = r cos u and
G y = r sin u, the Jacobian is

Jsr, ud = 4 4 = ` cos u
0x 0x
` = rscos2 u + sin2 ud = r.
0r 0u -r sin u
0y 0y sin u r cos u
r
0r 0u
0 1
Since we assume r Ú 0 when integrating in polar coordinates, ƒ J(r, u) ƒ = ƒ r ƒ = r, so that
Cartesian r u-plane
Equation (1) gives
x 5 r cos u

6 6
y 5 r sin u ƒsx, yd dx dy = ƒsr cos u, r sin ud r dr du. (3)
R G
y
This is the same formula we derived independently using a geometric argument for polar
p area in Section 15.4.
u5
2
Notice that the integral on the right-hand side of Equation (3) is not the integral of
1
ƒsr cos u, r sin ud over a region in the polar coordinate plane. It is the integral of the prod-
R
uct of ƒsr cos u, r sin ud and r over a region G in the Cartesian ru-plane.
Here is an example of a substitution in which the image of a rectangle under the coor-
u50 dinate transformation is a trapezoid. Transformations like this one are called linear trans-
x
0 1 formations.
Cartesian xy-plane
EXAMPLE 2 Evaluate
FIGURE 15.54 The equations x = x = sy>2d + 1
4 2x - y
L0 Lx = y>2
r cos u, y = r sin u transform G into R. dx dy
2

by applying the transformation

2x - y y
u = , y = (4)
2 2

and integrating over an appropriate region in the uy-plane.

Solution We sketch the region R of integration in the xy-plane and identify its bound-
aries (Figure 15.55).

v y

v52 y54
2 4

x5u1v y 5 2x
y 5 2v
u50 u51 R
G y 5 2x 2 2

u x
0 v50 1 0 1
y50

FIGURE 15.55 The equations x = u + y and y = 2y transform G into

R. Reversing the transformation by the equations u = s2x - yd>2 and
y = y>2 transforms R into G (Example 2).
 15.8 Substitutions in Multiple Integrals 907

To apply Equation (1), we need to find the corresponding uy-region G and the Jaco-
bian of the transformation. To find them, we first solve Equations (4) for x and y in terms
of u and y. From those equations it is easy to see that
x = u + y, y = 2y. (5)

We then find the boundaries of G by substituting these expressions into the equations for
the boundaries of R (Figure 15.55).

xy-equations for Corresponding uY-equations Simplified

the boundary of R for the boundary of G uY-equations

x = y>2 u + y = 2y>2 = y u = 0
x = s y>2d + 1 u + y = s2y>2d + 1 = y + 1 u = 1
y = 0 2y = 0 y = 0
y = 4 2y = 4 y = 2

The Jacobian of the transformation (again from Equations (5)) is

Jsu, yd = 4 4 = 4 4 = `1
0x 0x 0 0
` = 2.
su + yd su + yd
0u 0y 0u 0y 1
0y 0y 0 0 0 2
s2yd s2yd
0u 0y 0u 0y

We now have everything we need to apply Equation (1):

4 x = sy>2d + 1 2x - y y=2 u=1

L0 Lx = y>2 Ly = 0 Lu = 0
dx dy = u ƒ Jsu, yd ƒ du dy
2

cu 2 d dy =
v
2 1 2 1 2

L0 L0 L0 L0
v5u
1 = suds2d du dy = dy = 2.
0
u
0 G 1
u51
EXAMPLE 3 Evaluate
v 5 –2u

2x + y s y - 2xd2 dy dx.
1 1-x

L0 L0
–2
u v
x5 2
3 3
y5
2u v
1
Solution We sketch the region R of integration in the xy-plane and identify its boundaries
3 3 (Figure 15.56). The integrand suggests the transformation u = x + y and y = y - 2x.
y
Routine algebra produces x and y as functions of u and y:
1 u y 2u y
x = - , y = + . (6)
3 3 3 3
x1y51 From Equations (6), we can find the boundaries of the uy-region G (Figure 15.56).
x50
R
xy-equations for Corresponding uY-equations Simplified
x the boundary of R for the boundary of G uY-equations
0 1

a - b + a + b = 1
y50
u y 2u y
FIGURE 15.56 The equations x = x + y = 1 u = 1
3 3 3 3
su>3d - sy>3d and y = s2u>3d + sy>3d u y
transform G into R. Reversing the x = 0 - = 0 y = u
3 3
transformation by the equations u = x + y
2u y
and y = y - 2x transforms R into G y = 0 + = 0 y = - 2u
3 3
(Example 3).
908 Chapter 15: Multiple Integrals

The Jacobian of the transformation in Equations (6) is

Jsu, yd = 4 4 = 4 4 = 1.
0x 0x 1 1
-
0u 0y 3 3
0y 0y 2 1 3
0u 0y 3 3
Applying Equation (1), we evaluate the integral:

2x + y s y - 2xd2 dy dx =
1 1-x u=1 y=u

L0 L0 Lu = 0 Ly = -2u
u 1>2 y 2 ƒ Jsu, yd ƒ dy du

u 1>2 y 2 a b dy du = u 1>2 c y 3 d
1 u 1 y=u

LL L
1 1 1
= du
0 -2u 3 3 0 3 y = -2u

u 7>2 du = u 9/2 d = .
1 1 1

9L0 L
1 2 2
= u 1>2su 3 + 8u 3 d du =
0 9 0 9

In the next example we illustrate a nonlinear transformation of coordinates resulting from

simplifying the form of the integrand. Like the polar coordinates’ transformation, nonlinear
transformations can map a straight line boundary of a region into a curved boundary (or vice
versa with the inverse transformation). In general, nonlinear transformations are more complex
to analyze than linear ones, and a complete treatment is left to a more advanced course.
y
y5x EXAMPLE 4 Evaluate the integral
2
2 y y 1xy
A
y52

L1 L1>y
R
x e dx dy.

gration by substituting u = 2xy and y = 2y>x. Squaring these equations, we readily

Solution The square root terms in the integrand suggest that we might simplify the inte-
xy 5 1

have u 2 = xy and y2 = y>x, which imply that u 2y2 = y 2 and u 2>y2 = x 2. So we obtain
x
0 1 2 the transformation (in the same ordering of the variables as discussed before)
u
FIGURE 15.57 The region of x = y and y = uy.
integration R in Example 4.
Let’s first see what happens to the integrand itself under this transformation. The Jacobian
of the transformation is

Jsu, yd = 4 4 = † y2 †
y 0x 0x 1 -u
0u 0y y
2u
2 uy 5 2 ⇔ y 5 2 = y.
u 5 1 ⇔ xy 5 1
0y 0y y u
0u 0y
G
1 If G is the region of integration in the uy-plane, then by Equation (1) the transformed
y51⇔y5x double integral under the substitution is

y 1xy
6 A 6 6
u 2u
0 1 2 x e dx dy = ye u y du dy = 2ue u du dy.
R G G
FIGURE 15.58 The boundaries of the
region G correspond to those of region R The transformed integrand function is easier to integrate than the original one, so we
in Figure 15.57. Notice as we move proceed to determine the limits of integration for the transformed integral.

15.57. From the substitution equations u = 2xy and y = 2y>x, we see that the image of
counterclockwise around the region R, we The region of integration R of the original integral in the xy-plane is shown in Figure
also move counterclockwise around the

equations u = 1xy, y = 1y>x produce

region G. The inverse transformation the left-hand boundary xy = 1 for R is the vertical line segment u = 1, 2 Ú y Ú 1, in G
(see Figure 15.58). Likewise, the right-hand boundary y = x of R maps to the horizontal
the region G from the region R. line segment y = 1, 1 … u … 2, in G. Finally, the horizontal top boundary y = 2 of R