Sie sind auf Seite 1von 21

14.

3  Partial Derivatives 809

takes on all values from 0 to 1 regardless of how small 0 r 0 is, so that Using the Limit Definition
lim(x, y)S(0,0) x2 > ( x2 + y 2 ) does not exist. Each of Exercises 73–78 gives a function ƒ(x, y) and a positive num-
In each of these instances, the existence or nonexistence of the ber e. In each exercise, show that there exists a d 7 0 such that for all
limit as r S 0 is fairly clear. Shifting to polar coordinates does not (x, y),
always help, however, and may even tempt us to false conclusions. 2x2 + y 2 6 d 1 0 ƒ(x, y) - ƒ(0, 0) 0 6 e.
For example, the limit may exist along every straight line (or ray)
u = constant and yet fail to exist in the broader sense. Example 5 73. ƒ(x, y) = x2 + y 2, e = 0.01
illustrates this point. In polar coordinates, ƒ(x, y) = ( 2x2y ) > ( x4 + y 2 ) 74. ƒ(x, y) = y> ( x2 + 1 ) , e = 0.05
becomes
75. ƒ(x, y) = (x + y)> ( x2 + 1 ) , e = 0.01
r cos u sin 2u 76. ƒ(x, y) = (x + y)>(2 + cos x), e = 0.02
ƒ(r cos u, r sin u) = 2
r cos4 u + sin2 u xy 2
77. ƒ(x, y) = 2 and ƒ(0, 0) = 0, e = 0.04
for r ≠ 0. If we hold u constant and let r S 0, the limit is 0. On the x + y2
path y = x2, however, we have r sin u = r 2 cos2 u and
x3 + y 4
r cos u sin 2u 78. ƒ(x, y) = and ƒ(0, 0) = 0, e = 0.02
ƒ(r cos u, r sin u) = 2 x2 + y 2
r cos4 u + (r cos2 u)2
Each of Exercises 79–82 gives a function ƒ(x, y, z) and a positive
2r cos2 u sin u r sin u number e. In each exercise, show that there exists a d 7 0 such that
= = 2 = 1.
2r 2 cos4 u r cos2 u for all (x, y, z),
In Exercises 65–70, find the limit of ƒ as (x, y) S (0, 0) or show that 2x2 + y 2 + z2 6 d 1 0 ƒ(x, y, z) - ƒ(0, 0, 0) 0 6 e.
the limit does not exist.
x3 - xy 2 x3 - y 3 79. ƒ(x, y, z) = x2 + y 2 + z2, e = 0.015
65. ƒ(x, y) = 2 66. ƒ(x, y) = cos a 2 b 80. ƒ(x, y, z) = xyz, e = 0.008
x + y 2
x + y2
2 x + y + z
y 2x 81. ƒ(x, y, z) = 2 , e = 0.015
67. ƒ(x, y) = 68. ƒ(x, y) = x + y 2 + z2 + 1
x + y2
2
x2 + x + y 2
0x0 + 0y0 82. ƒ(x, y, z) = tan2 x + tan2 y + tan2 z, e = 0.03
69. ƒ(x, y) = tan-1 a b 83. Show that ƒ(x, y, z) = x + y - z is continuous at every point
x2 + y 2
(x0 , y0 , z0).
x2 - y 2
70. ƒ(x, y) = 84. Show that ƒ(x, y, z) = x2 + y 2 + z2 is continuous at the origin.
x2 + y 2

In Exercises 71 and 72, define ƒ(0, 0) in a way that extends ƒ to be


continuous at the origin.
3x2 - x2y 2 + 3y 2
71. ƒ(x, y) = ln a b
x2 + y 2
3x2y
72. ƒ(x, y) =
x + y2
2

14.3 Partial Derivatives


The calculus of several variables is similar to single-variable calculus applied to several
variables one at a time. When we hold all but one of the independent variables of a func-
tion constant and differentiate with respect to that one variable, we get a “partial” deriva-
tive. This section shows how partial derivatives are defined and interpreted geometrically,
and how to calculate them by applying the rules for differentiating functions of a single
variable. The idea of differentiability for functions of several variables requires more than
the existence of the partial derivatives because a point can be approached from many dif-
ferent directions. However, we will see that differentiable functions of several variables
behave similarly to differentiable single-variable functions. In particular, they are continu-
ous and can be well approximated by linear functions.

Partial Derivatives of a Function of Two Variables


If (x0, y0) is a point in the domain of a function ƒ(x, y), the vertical plane y = y0 will cut
the surface z = ƒ(x, y) in the curve z = ƒ(x, y0) (Figure 14.16). This curve is the graph of
810 Chapter 14 Partial Derivatives

Vertical axis in
the plane y = y0
P(x0, y0, f(x0, y0))
z = f (x, y)
The curve z = f(x, y0)
in the plane y = y0

Tangent line

0
x0
y0
x
(x0, y0) y
(x0 + h, y0)
Horizontal axis in the plane y = y0

FIGURE 14.16 The intersection of the plane y = y0


with the surface z = ƒ(x, y), viewed from above the first
quadrant of the xy-plane.

the function z = ƒ(x, y0) in the plane y = y0 . The horizontal coordinate in this plane is x;
the vertical coordinate is z. The y-value is held constant at y0 , so y is not a variable.
We define the partial derivative of ƒ with respect to x at the point (x0 , y0) as the ordi-
nary derivative of ƒ(x, y0) with respect to x at the point x = x0 . To distinguish partial
derivatives from ordinary derivatives we use the symbol 0 rather than the d previously
used. In the definition, h represents a real number, positive or negative.

DEFINITION The partial derivative of ƒ(x, y) with respect to x at the point


(x0 , y0) is
0ƒ ƒ(x0 + h, y0) - ƒ(x0 , y0)
2 = lim ,
0x (x , y ) h S 0 h
0 0

provided the limit exists.

The partial derivative of ƒ(x, y) with respect to x at the point (x0, y0) is the same as the
ordinary derivative of ƒ(x, y0) at the point x0:
0ƒ d
2 = ƒ(x, y0) 2 .
0x (x , y ) dx x = x0
0 0

A variety of notations are used to denote the partial derivative at a point (x0, y0),
including
0ƒ 0z 2
(x , y ), ƒx(x0, y0), and .
0x 0 0 0x (x , y )
0 0

When we do not specify a specific point (x0, y0) at which the partial derivative is being
evaluated, then the partial derivative becomes a function whose domain is the points where
the partial derivative exists. Notations for this function include
0ƒ 0z
, ƒx , and .
0x 0x
14.3  Partial Derivatives 811

The slope of the curve z = ƒ(x, y0) at the point P(x0 , y0 , ƒ(x0 , y0)) in the plane y = y0
is the value of the partial derivative of ƒ with respect to x at (x0 , y0). (In Figure 14.16 this
slope is negative.) The tangent line to the curve at P is the line in the plane y = y0 that
passes through P with this slope. The partial derivative 0ƒ>0x at (x0 , y0) gives the rate of
change of ƒ with respect to x when y is held fixed at the value y0.
The definition of the partial derivative of ƒ(x, y) with respect to y at a point (x0 , y0) is
similar to the definition of the partial derivative of ƒ with respect to x. We hold x fixed at
the value x0 and take the ordinary derivative of ƒ(x0 , y) with respect to y at y0.

DEFINITION The partial derivative of ƒ(x, y) with respect to y at the point


(x0 , y0) is
0ƒ d ƒ(x0 , y0 + h) - ƒ(x0 , y0)
2 = ƒ(x0 , y) 2 = lim ,
0y (x , y ) dy y = y0 h S 0 h
0 0

provided the limit exists.

Vertical axis The slope of the curve z = ƒ(x0 , y) at the point P(x0 , y0 , ƒ(x0 , y0)) in the vertical
in the plane z plane x = x0 (Figure 14.17) is the partial derivative of ƒ with respect to y at (x0 , y0). The
x = x0
tangent line to the curve at P is the line in the plane x = x0 that passes through P with this
slope. The partial derivative gives the rate of change of ƒ with respect to y at (x0, y0) when
x is held fixed at the value x0 .
Tangent line The partial derivative with respect to y is denoted the same way as the partial deriva-
P(x0, y0, f (x0, y0))
tive with respect to x:
0ƒ 0ƒ
z = f (x, y) (x , y ), ƒy(x0 , y0), , ƒy .
0y 0 0 0y
Notice that we now have two tangent lines associated with the surface z = ƒ(x, y) at
0
x0 y0
the point P(x0, y0, ƒ(x0, y0)) (Figure 14.18). Is the plane they determine tangent to the sur-
face at P? We will see that it is for the differentiable functions defined at the end of this
x section, and we will learn how to find the tangent plane in Section 14.6. First we have to
(x0, y0) y better understand partial derivatives.
(x0, y0 + k)
z
The curve z = f (x0, y)
Horizontal axis
in the plane
in the plane x = x0 This tangent line
x = x0 P(x0, y0, f (x0, y0))
has slope fy (x0, y0). This tangent line
FIGURE 14.17 The intersection of the has slope fx(x0, y0).
plane x = x0 with the surface z = ƒ(x, y), The curve z = f(x0, y) in
the vertical plane x = x0 The curve z = f (x, y0) in
viewed from above the first quadrant of
the vertical plane y = y0
the xy-plane.

z = f (x, y)

y = y0 (x0, y0) x = x0
y

FIGURE 14.18 Figures 14.16 and 14.17 combined. The tangent


lines at the point (x0 , y0 , ƒ(x0 , y0)) determine a plane that, in this
picture at least, appears to be tangent to the surface.
812 Chapter 14 Partial Derivatives

Calculations
The definitions of 0ƒ>0x and 0ƒ>0y give us two different ways of differentiating ƒ at a
point: with respect to x in the usual way while treating y as a constant and with respect to y
in the usual way while treating x as a constant. As the following examples show, the values
of these partial derivatives are usually different at a given point (x0, y0).

EXAMPLE 1 Find the values of 0ƒ>0x and 0ƒ>0y at the point (4, -5) if
ƒ(x, y) = x2 + 3xy + y - 1.

Solution To find 0ƒ>0x, we treat y as a constant and differentiate with respect to x:



(x + 3xy + y - 1) = 2x + 3 # 1 # y + 0 - 0 = 2x + 3y.
0 2
=
0x 0x
The value of 0ƒ>0x at (4, -5) is 2(4) + 3(-5) = -7.
To find 0ƒ>0y, we treat x as a constant and differentiate with respect to y:

0y
=
0y
( x + 3xy + y - 1 ) = 0 + 3 # x # 1 + 1 - 0 = 3x + 1.
0 2

The value of 0ƒ>0y at (4, -5) is 3(4) + 1 = 13.

EXAMPLE 2 Find 0ƒ>0y as a function if ƒ(x, y) = y sin xy.

Solution We treat x as a constant and ƒ as a product of y and sin xy:


0ƒ 0 0 0
= ( y sin xy) = y sin xy + (sin xy) ( y)
0y 0y 0y 0y
0
= ( y cos xy) (xy) + sin xy = xy cos xy + sin xy.
0y

EXAMPLE 3 Find ƒx and ƒy as functions if


2y
ƒ(x, y) = .
y + cos x

Solution We treat ƒ as a quotient. With y held constant, we use the quotient rule to get

0 0
( y + cos x) (2y) - 2y ( y + cos x)
0 2y 0x 0x
ƒx = a b =
0x y + cos x ( y + cos x)2
( y + cos x)(0) - 2y(-sin x) 2y sin x
= = .
( y + cos x)2
( y + cos x)2

With x held constant and again applying the quotient rule, we get

0 0
( y + cos x) (2y) - 2y ( y + cos x)
0 2y 0y dy
ƒy = a b =
0y y + cos x ( y + cos x)2
( y + cos x)(2) - 2y(1) 2 cos x
= = .
( y + cos x)2 ( y + cos x)2

Implicit differentiation works for partial derivatives the way it works for ordinary
derivatives, as the next example illustrates.
14.3  Partial Derivatives 813

EXAMPLE 4 Find 0z>0x assuming that the equation


yz - ln z = x + y
defines z as a function of the two independent variables x and y and the partial derivative
exists.

Solution We differentiate both sides of the equation with respect to x, holding y constant
and treating z as a differentiable function of x:
0 0 0x 0y
( yz) - ln z = +
0x 0x 0x 0x
0z 1 0z 0 0z
y - z = 1 + 0 With y constant, ( yz) = y .
0x 0x 0x 0x

1 0z
ay - z b = 1
0x
0z z
= .
0x yz - 1

z
EXAMPLE 5 The plane x = 1 intersects the paraboloid z = x2 + y 2 in a parabola.
Plane
Find the slope of the tangent to the parabola at (1, 2, 5) (Figure 14.19).
x=1
Surface
z = x2 + y2
Solution The parabola lies in a plane parallel to the yz-plane, and the slope is the value
of the partial derivative 0z>0y at (1, 2):
0z 2 0 2
Tangent =
0y (1,2) 0y
( x + y2 ) 2 = 2y 2 = 2(2) = 4.
line (1,2) (1,2)
(1, 2, 5)
As a check, we can treat the parabola as the graph of the single-variable function
z = (1)2 + y 2 = 1 + y 2 in the plane x = 1 and ask for the slope at y = 2. The slope,
calculated now as an ordinary derivative, is
dz 2 d
= ( 1 + y2 ) 2 = 2y 2 = 4.
dy y = 2 dy y=2 y=2
1 2
y
x=1
x
Functions of More Than Two Variables
FIGURE 14.19 The tangent to the curve
of intersection of the plane x = 1 and The definitions of the partial derivatives of functions of more than two independent vari-
surface z = x2 + y 2 at the point (1, 2, 5) ables are similar to the definitions for functions of two variables. They are ordinary deriva-
(Example 5). tives with respect to one variable, taken while the other independent variables are held
constant.

EXAMPLE 6 If x, y, and z are independent variables and


ƒ(x, y, z) = x sin ( y + 3z),
then
0ƒ 0 0
= 3 x sin ( y + 3z)4 = x sin ( y + 3z) x held constant
0z 0z 0z
0
= x cos ( y + 3z) ( y + 3z) Chain rule
0z
= 3x cos ( y + 3z). y held constant
814 Chapter 14 Partial Derivatives

R1 EXAMPLE 7 If resistors of R1, R2, and R3 ohms are connected in parallel to make an
R-ohm resistor, the value of R can be found from the equation
R2
1 1 1 1
R3 = + +
R R1 R2 R3

(Figure 14.20). Find the value of 0R>0R2 when R1 = 30, R2 = 45, and R3 = 90 ohms.

Solution To find 0R>0R2, we treat R1 and R3 as constants and, using implicit differentia-
+ −
tion, differentiate both sides of the equation with respect to R2:
FIGURE 14.20 Resistors arranged this
0 1 0 1 1 1
way are said to be connected in parallel a b = a + + b
0R2 R 0R2 R1 R2 R3
(Example 7). Each resistor lets a portion
of the current through. Their equivalent 1 0R 1
- = 0 - + 0
resistance R is calculated with the formula R2 0R2 R2 2
1 1 1 1
= + + . 0R R2 R 2
R R1 R2 R3 = = a b .
0R2 R2 2 R2

When R1 = 30, R2 = 45, and R3 = 90,


1 1 1 1 3 + 2 + 1 6 1
= + + = = = ,
R 30 45 90 90 90 15

so R = 15 and
0R 15 2 1 2 1
= a b = a b = .
0R2 45 3 9

Thus at the given values, a small change in the resistance R2 leads to a change in R about
1>9th as large.

Partial Derivatives and Continuity


A function ƒ(x, y) can have partial derivatives with respect to both x and y at a point
without the function being continuous there. This is different from functions of a single
variable, where the existence of a derivative implies continuity. If the partial derivatives of
0, xy ≠ 0 ƒ(x, y) exist and are continuous throughout a disk centered at (x0, y0), however, then ƒ is
z=
1, xy = 0 z continuous at (x0 , y0), as we see at the end of this section.
L1
1
EXAMPLE 8 Let
0 0, xy ≠ 0
ƒ(x, y) = e
L2 1, xy = 0
x
y (Figure 14.21).
(a) Find the limit of ƒ as (x, y) approaches (0, 0) along the line y = x.
FIGURE 14.21 The graph of (b) Prove that ƒ is not continuous at the origin.
0, xy ≠ 0 (c) Show that both partial derivatives 0ƒ>0x and 0ƒ>0y exist at the origin.
ƒ(x, y) = e
1, xy = 0
Solution
consists of the lines L 1 and L 2 (lying
1 unit above the xy-plane) and the four (a) Since ƒ(x, y) is constantly zero along the line y = x (except at the origin), we have
open quadrants of the xy-plane. The
function has partial derivatives at the origin lim ƒ(x, y) 2 = lim 0 = 0.
but is not continuous there (Example 8). (x, y) S (0,0) y=x (x, y) S (0,0)
14.3  Partial Derivatives 815

(b) Since ƒ(0, 0) = 1, the limit in part (a) is not equal to ƒ(0, 0), which proves that ƒ is not
continuous at (0, 0).
(c) To find 0ƒ>0x at (0, 0), we hold y fixed at y = 0. Then ƒ(x, y) = 1 for all x, and the
graph of ƒ is the line L 1 in Figure 14.21. The slope of this line at any x is 0ƒ>0x = 0.
In particular, 0ƒ>0x = 0 at (0, 0). Similarly, 0ƒ>0y is the slope of line L 2 at any y, so
0ƒ>0y = 0 at (0, 0).

What Example 8 suggests is that we need a stronger requirement for differentiability


in higher dimensions than the mere existence of the partial derivatives. We define differ-
entiability for functions of two variables (which is slightly more complicated than for
single-variable functions) at the end of this section and then revisit the connection to
continuity.

Second-Order Partial Derivatives


When we differentiate a function ƒ(x, y) twice, we produce its second-order derivatives.
These derivatives are usually denoted by
0 2ƒ 0 2ƒ
or ƒxx , or ƒyy ,
0x2 0y 2
0 2ƒ 0 2ƒ
or ƒyx , and or ƒxy .
0x 0y 0y 0x
The defining equations are
0 2ƒ 0 0ƒ 0 2ƒ 0 0ƒ
= a b, = a b,
0x 2 0x 0x 0x 0y 0x 0y
and so on. Notice the order in which the mixed partial derivatives are taken:
0 2ƒ
Differentiate first with respect to y, then with respect to x.
0x 0y
ƒyx = (ƒy)x Means the same thing.

HISTORICAL BIOGRAPHY EXAMPLE 9 If ƒ(x, y) = x cos y + yex, find the second-order derivatives
Pierre-Simon Laplace 0 2ƒ 0 2ƒ 0 2ƒ 0 2ƒ
(1749–1827) , , , and .
0x2 0y 0x 0y 2 0x 0y
www.goo.gl/2PxgfB

Solution The first step is to calculate both first partial derivatives.


0ƒ 0 0ƒ 0
= (x cos y + yex) = (x cos y + yex)
0x 0x 0y 0y
= cos y + yex = -x sin y + ex
Now we find both partial derivatives of each first partial:
0 2ƒ 0 0ƒ 0 2ƒ 0 0ƒ
= a b = -sin y + ex = a b = -sin y + ex
0y 0x 0y 0x 0x 0y 0x 0y
0 2ƒ 0 0ƒ 0 2ƒ 0 0ƒ
= a b = yex. = a b = -x cos y.
0x 2 0x 0x 0y 2 0y 0y
816 Chapter 14 Partial Derivatives

The Mixed Derivative Theorem


You may have noticed that the “mixed” second-order partial derivatives
0 2ƒ 0 2ƒ
and
0y 0x 0x 0y
in Example 9 are equal. This is not a coincidence. They must be equal whenever
ƒ, ƒx , ƒy , ƒxy , and ƒyx are continuous, as stated in the following theorem. However, the
mixed derivatives can be different when the continuity conditions are not satisfied (see
Exercise 82).

THEOREM 2—The Mixed Derivative Theorem


If ƒ(x, y) and its partial derivatives ƒx , ƒy , ƒxy , and ƒyx are defined throughout an
open region containing a point (a, b) and are all continuous at (a, b), then
ƒxy(a, b) = ƒyx(a, b).

HISTORICAL BIOGRAPHY
Theorem 2 is also known as Clairaut’s Theorem, named after the French mathemati-
Alexis Clairaut
cian Alexis Clairaut, who discovered it. A proof is given in Appendix 9. Theorem 2 says
(1713–1765) that to calculate a mixed second-order derivative, we may differentiate in either order,
www.goo.gl/MYhzm3 provided the continuity conditions are satisfied. This ability to proceed in different order
sometimes simplifies our calculations.

0 2w
EXAMPLE 10 Find if
0x 0y

ey
w = xy + .
y + 1
2

Solution The symbol 0 2w>0x 0y tells us to differentiate first with respect to y and then
with respect to x. However, if we interchange the order of differentiation and differentiate
first with respect to x we get the answer more quickly. In two steps,
0w 0 2w
= y and = 1.
0x 0y 0x
If we differentiate first with respect to y, we obtain 0 2w>0x 0y = 1 as well. We can differen-
tiate in either order because the conditions of Theorem 2 hold for w at all points (x0 , y0 ).

Partial Derivatives of Still Higher Order


Although we will deal mostly with first- and second-order partial derivatives, because
these appear the most frequently in applications, there is no theoretical limit to how many
times we can differentiate a function as long as the derivatives involved exist. Thus, we get
third- and fourth-order derivatives denoted by symbols like
0 3ƒ
= ƒyyx ,
0x 0y 2
0 4ƒ
= ƒyyxx ,
0x2 0y 2
and so on. As with second-order derivatives, the order of differentiation is immaterial as
long as all the derivatives through the order in question are continuous.
14.3  Partial Derivatives 817

EXAMPLE 11 Find ƒyxyz if ƒ(x, y, z) = 1 - 2xy 2z + x2y.

Solution We first differentiate with respect to the variable y, then x, then y again, and
finally with respect to z:
ƒy = -4xyz + x2
ƒyx = -4yz + 2x
ƒyxy = -4z
ƒyxyz = -4.

Differentiability
The concept of differentiability for functions of several variables is more complicated than
for single-variable functions because a point in the domain can be approached along more
than one path. In defining the partial derivatives for a function of two variables, we inter-
sected the surface of the graph with vertical planes parallel to the xz- and yz-planes, creat-
ing a curve on each plane, called a trace. The partial derivatives were seen as the slopes of
the two tangent lines to these trace curves at the point on the surface corresponding to the
point (x0 , y0) being approached in the domain. (See Figure 14.18.) For a differentiable
function, it would seem reasonable to assume that if we were to rotate slightly one of these
vertical planes, keeping it vertical but no longer parallel to its coordinate plane, then a
smooth trace curve would appear on that plane that would have a tangent line at the point
on the surface having a slope differing just slightly from what it was before (when the
plane was parallel to its coordinate plane). However, the mere existence of the original
partial derivative does not guarantee that result. Just as having a limit in the x- and y-
coordinate directions alone does not imply the function itself has a limit at (x0, y0), as we
saw in Figure 14.21, so is it the case that the existence of both partial derivatives is not
enough by itself to ensure derivatives exist for trace curves in other vertical planes. For the
existence of differentiability, a property is needed to ensure that no abrupt change occurs
in the function resulting from small changes in the independent variables along any path
approaching (x0 , y0), paths along which both variables x and y are allowed to change,
rather than just one of them at a time.
In our study of functions of a single variable, we found that if a function y = ƒ(x) is
differentiable at x = x0, then the change ∆y resulting in the change of x from
x0 to x0 + ∆x is close to the change ∆L along the tangent line (or linear approximation L
of the function ƒ at x0 ). That is, from Equation (1) in Section 3.9,
∆y = ƒ′(x0)∆x + e ∆x
in which e S 0 as ∆x S 0. The extension of this result is what we use to define differen-
tiability for functions of two variables.

DEFINITION A function z = ƒ(x, y) is differentiable at (x0 , y0) if ƒx(x0 , y0)


and ƒy(x0 , y0) exist and ∆z = ƒ(x0 + ∆x, y0 + ∆y) - ƒ(x0, y0) satisfies an equa-
tion of the form
∆z = ƒx(x0 , y0)∆x + ƒy(x0 , y0)∆y + e1 ∆x + e2 ∆y
in which each of e1, e2 S 0 as both ∆x, ∆y S 0. We call ƒ differentiable if it is
differentiable at every point in its domain, and say that its graph is a smooth surface.

The following theorem (proved in Appendix 9) and its accompanying corollary tell us
that functions with continuous first partial derivatives at (x0 , y0) are differentiable there,
and they are closely approximated locally by a linear function. We study this approxima-
tion in Section 14.6.
14.3  Partial Derivatives 819

28. ƒ(x, y, z) = sec-1 (x + yz) 62. The fifth-order partial derivative 0 5ƒ>0x20y 3 is zero for each of the
29. ƒ(x, y, z) = ln (x + 2y + 3z) following functions. To show this as quickly as possible, which
variable would you differentiate with respect to first: x or y? Try to
30. ƒ(x, y, z) = yz ln (xy)
answer without writing anything down.
31. ƒ(x, y, z) = e-(x + y2 + z2)
2

a. ƒ(x, y) = y 2x4ex + 2
32. ƒ(x, y, z) = e-xyz
b. ƒ(x, y) = y 2 + y ( sin x - x4 )
33. ƒ(x, y, z) = tanh (x + 2y + 3z)
c. ƒ(x, y) = x2 + 5xy + sin x + 7ex
34. ƒ(x, y, z) = sinh (xy - z2)
d. ƒ(x, y) = xey >2
2

In Exercises 35–40, find the partial derivative of the function with Using the Partial Derivative Definition
respect to each variable. In Exercises 63–66, use the limit definition of partial derivative to
compute the partial derivatives of the functions at the specified points.
35. ƒ(t, a) = cos (2pt - a)
0ƒ 0ƒ
36. g(u, y) = y2e(2u>y) 63. ƒ(x, y) = 1 - x + y - 3x2y, and at (1, 2)
0x 0y
37. h(r, f, u) = r sin f cos u
0ƒ 0ƒ
38. g(r, u, z) = r(1 - cos u) - z 64. ƒ(x, y) = 4 + 2x - 3y - xy 2, and at (-2, 1)
0x 0y
39. Work done by the heart (Section 3.9, Exercise 51)
0ƒ 0ƒ
65. ƒ(x, y) = 22x + 3y - 1, and at (-2, 3)
Vdy2 0x 0y
W(P, V, d, y, g) = PV +
2g
sin ( x3 + y 4 )
, (x, y) ≠ (0, 0)
40. Wilson lot size formula (Section 4.5, Exercise 57) 66. ƒ(x, y) = c x2 + y 2
km hq 0, (x, y) = (0, 0),
A(c, h, k, m, q) = q + cm +
2 0ƒ 0ƒ
and at (0, 0)
0x 0y
Calculating Second-Order Partial Derivatives 67. Let ƒ(x, y) = 2x + 3y - 4. Find the slope of the line tangent to
Find all the second-order partial derivatives of the functions in Exer- this surface at the point (2, - 1) and lying in the a. plane x = 2
cises 41–50. b. plane y = - 1.
41. ƒ(x, y) = x + y + xy 42. ƒ(x, y) = sin xy 68. Let ƒ(x, y) = x2 + y 3. Find the slope of the line tangent to this
43. g(x, y) = x2y + cos y + y sin x surface at the point (-1, 1) and lying in the a. plane x = - 1
44. h(x, y) = xey + y + 1 45. r(x, y) = ln (x + y) b. plane y = 1.
46. s(x, y) = tan-1 ( y>x) 47. w = x2 tan (xy) 69. Three variables Let w = ƒ(x, y, z) be a function of three in-
x2 - y dependent variables and write the formal definition of the partial
48. w = ye 49. w = x sin ( x y )
2
derivative 0ƒ>0z at (x0 , y0 , z0). Use this definition to find 0ƒ>0z at
x - y (1, 2, 3) for ƒ(x, y, z) = x2yz2.
50. w = 2 51. ƒ(x, y) = x2y 3 - x4 + y 5
x + y 70. Three variables Let w = ƒ(x, y, z) be a function of three in-
52. g(x, y) = cos x2 - sin 3y 53. z = x sin (2x - y 2) dependent variables and write the formal definition of the partial
derivative 0ƒ>0y at (x0 , y0 , z0). Use this definition to find 0ƒ>0y at
54. z = xe x>y
2

(- 1, 0, 3) for ƒ(x, y, z) = - 2xy 2 + yz2.


Mixed Partial Derivatives In Exercises 71–74, find a function z = ƒ(x, y) whose partial deriva-
In Exercises 55–60, verify that wxy = wyx . tives are as given, or explain why this is impossible.
55. w = ln (2x + 3y) 56. w = ex + x ln y + y ln x 0ƒ 0ƒ
71. = 3x2y 2 - 2x, = 2x3y + 6y
57. w = xy + x y + x y
2 2 3 3 4 0x 0y
58. w = x sin y + y sin x + xy 0ƒ 0ƒ
= 2xe xy + x2y 2e xy + 3, = 2x3ye xy - ey
2 2 2
72.
2 3x - y 0x 0y
x
59. v = 60. v = 0ƒ 0ƒ
y3 x + y 2y 2x
73. = , =
0x (x + y) 2 0y (x + y)2
61. Which order of differentiation will calculate fxy faster: x first or y
first? Try to answer without writing anything down. 0ƒ 0ƒ
74. = xy cos (xy) + sin (xy), = x cos (xy)
0x 0y
a. ƒ(x, y) = x sin y + ey
Differentiating Implicitly
b. ƒ(x, y) = 1>x
75. Find the value of 0z>0x at the point (1, 1, 1) if the equation
c. ƒ(x, y) = y + (x>y)
xy + z3x - 2yz = 0
d. ƒ(x, y) = y + x2y + 4y 3 - ln ( y 2 + 1)
e. ƒ(x, y) = x2 + 5xy + sin x + 7ex defines z as a function of the two independent variables x and y and
f. ƒ(x, y) = x ln xy the partial derivative exists.
820 Chapter 14 Partial Derivatives

76. Find the value of 0x>0z at the point (1, -1, - 3) if the equation obtained by dropping the 0 2ƒ>0z2 term from the previous
xz + y ln x - x2 + 4 = 0 equation, describes potentials and steady-state temperature
distributions in a plane (see the accompanying figure). The plane (a)
defines x as a function of the two independent variables y and z and may be treated as a thin slice of the solid (b) perpendicular to the
the partial derivative exists. z-axis.
Exercises 77 and 78 are about the triangle shown here. '2f '2f
+ 2=0
'x 2 'y
B
c
a (a)
A
C b

77. Express A implicitly as a function of a, b, and c and calculate


0A>0a and 0A>0b.
78. Express a implicitly as a function of A, b, and B and calculate
0a>0A and 0a>0B.
79. Two dependent variables Express yx in terms of u and y if the
equations x = y ln u and y = u ln y define u and y as functions '2f '2f '2f
+ 2 + 2 =0
of the independent variables x and y, and if yx exists. (Hint: Dif- 'x 2
'y 'z
ferentiate both equations with respect to x and solve for yx by
eliminating ux .) (b)
80. Two dependent variables Find 0x>0u and 0y>0u if the equations
u = x2 - y 2 and y = x2 - y define x and y as functions of the in-
dependent variables u and y, and the partial derivatives exist. (See
the hint in Exercise 79.) Then let s = x2 + y 2 and find 0s>0u.

Theory and Examples


y 3, y Ú 0 Boundary temperatures controlled
81. Let ƒ(x, y) = e
- y 2, y 6 0.
Show that each function in Exercises 83–90 satisfies a Laplace
Find ƒx , ƒy , ƒxy , and ƒyx , and state the domain for each partial equation.
derivative. 83. ƒ(x, y, z) = x2 + y 2 - 2z2
x2 - y 2 84. ƒ(x, y, z) = 2z3 - 3 ( x2 + y 2 ) z
xy , if (x, y) ≠ 0, 85. ƒ(x, y) = e-2y cos 2x
82. Let ƒ(x, y) = c x2 + y 2
0, if (x, y) = 0. 86. ƒ(x, y) = ln 2x2 + y 2
The graph of ƒ is shown on page 799. 87. ƒ(x, y) = 3x + 2y - 4
x
0ƒ 0ƒ 88. ƒ(x, y) = tan-1 y
a. Show that (x, 0) = x for all x, and (0, y) = - y for all y.
0y 0x
89. ƒ(x, y, z) = ( x2 + y 2 + z2 ) -1>2
0 2ƒ 0 2ƒ
b. Show that (0, 0) ≠ (0, 0). 90. ƒ(x, y, z) = e3x + 4y cos 5z
0y 0x 0x 0y
The wave equation If we stand on an ocean shore and take a snap-
The three-dimensional Laplace equation shot of the waves, the picture shows a regular pattern of peaks and
0 2ƒ 0 2ƒ 0 2ƒ valleys in an instant of time. We see periodic vertical motion in space,
+ 2 + 2 = 0 with respect to distance. If we stand in the water, we can feel the rise
0x2 0y 0z
and fall of the water as the waves go by. We see periodic vertical
is satisfied by steady-state temperature distributions T = ƒ(x, y, z) motion in time. In physics, this beautiful symmetry is expressed by
in space, by gravitational potentials, and by electrostatic poten- the one-dimensional wave equation
tials. The two-dimensional Laplace equation
0 2w 0 2w
0 2ƒ 0 2ƒ = c2 2 ,
+ 2 = 0, 0t 2
0x
0x2 0y
14.4  The Chain Rule 821

where w is the wave height, x is the distance variable, t is the time 100. The heat equation An important partial differential equation
variable, and c is the velocity with which the waves are propagated. that describes the distribution of heat in a region at time t can be
w represented by the one-dimensional heat equation
0ƒ 0 2ƒ
= 2.
0t 0x
Show that u(x, t) = sin (ax) # e-bt satisfies the heat equation for
constants a and b. What is the relationship between a and b for
this function to be a solution?
xy 2
0 , (x, y) ≠ (0, 0)
101. Let ƒ(x, y) = c x2 + y 4
x
0, (x, y) = (0, 0).
In our example, x is the distance across the ocean’s surface, but
Show that ƒx(0, 0) and ƒy(0, 0) exist, but ƒ is not differentiable at
in other applications, x might be the distance along a vibrating string,
(0, 0). (Hint: Use Theorem 4 and show that ƒ is not continuous at
distance through air (sound waves), or distance through space (light
(0, 0).)
waves). The number c varies with the medium and type of wave.
Show that the functions in Exercises 91–97 are all solutions of 0, x2 6 y 6 2x2
102. Let ƒ(x, y) = b
the wave equation. 1, otherwise.
91. w = sin (x + ct) 92. w = cos (2x + 2ct) Show that ƒx(0, 0) and ƒy(0, 0) exist, but ƒ is not differentiable at
93. w = sin (x + ct) + cos (2x + 2ct) (0, 0).
94. w = ln (2x + 2ct) 95. w = tan (2x - 2ct) 103. The Korteweg-deVries equation
96. w = 5 cos (3x + 3ct) + ex + ct This nonlinear differential equation, which describes wave mo-
97. w = ƒ(u), where ƒ is a differentiable function of u, and u = tion on shallow water surfaces, is given by
a(x + ct), where a is a constant 4ut + uxxx + 12u ux = 0.
98. Does a function ƒ(x, y) with continuous first partial derivatives
Show that u(x, t) = sech2 (x - t) satisfies the Kortweg-deVries
throughout an open region R have to be continuous on R? Give
equation.
reasons for your answer. 1
104. Show that T = satisfies the equation Txx + Tyy = T 3.
99. If a function ƒ(x, y) has continuous second partial derivatives 2x2 + y 2
throughout an open region R, must the first-order partial deriva-
tives of ƒ be continuous on R? Give reasons for your answer.

14.4 The Chain Rule


The Chain Rule for functions of a single variable studied in Section 3.6 says that when
w = ƒ(x) is a differentiable function of x and x = g(t) is a differentiable function of t, w is
a differentiable function of t and dw > dt can be calculated by the formula
To find dw>dt, we read down the route
dw dw dx
from w to t, multiplying derivatives = .
along the way. dt dx dt
Chain Rule For this composite function w(t) = ƒ(g(t)), we can think of t as the independent variable
Dependent and x = g(t) as the “intermediate variable,” because t determines the value of x which in
w = f (x)
variable turn gives the value of w from the function ƒ. We display the Chain Rule in a “dependency
diagram” in the margin. Such diagrams capture which variables depend on which.
dw
dx
For functions of several variables the Chain Rule has more than one form, which
depends on how many independent and intermediate variables are involved. However,
Intermediate
x once the variables are taken into account, the Chain Rule works in the same way we just
variable
discussed.
dx
dt
Independent Functions of Two Variables
t variable
dw dw dx The Chain Rule formula for a differentiable function w = ƒ(x, y) when x = x(t) and
=
dt dx dt y = y(t) are both differentiable functions of t is given in the following theorem.
822 Chapter 14 Partial Derivatives

THEOREM 5—Chain Rule For Functions of One Independent Variable


and Two Intermediate Variables
If w = ƒ(x, y) is differentiable and if x = x(t), y = y(t) are differentiable func-
tions of t, then the composition w = ƒ(x(t), y(t)) is a differentiable function of t
and
dw
= ƒx(x(t), y(t)) x′(t) + ƒy(x(t), y(t)) y′(t),
dt
or
dw 0ƒ dx 0ƒ dy
= + .
dt 0x dt 0y dt

Proof The proof consists of showing that if x and y are differentiable at t = t0 , then w
0ƒ 0w
Each of , , ƒ indicates the partial
0x 0x x
is differentiable at t0 and
derivative of ƒ with respect to x. dw 0w dx 0w dy
(t ) = (P ) (t ) + (P ) (t ),
dt 0 0x 0 dt 0 0y 0 dt 0
where P0 = (x(t0), y(t0)).
Let ∆x, ∆y, and ∆w be the increments that result from changing t from t0 to t0 + ∆t.
Since ƒ is differentiable (see the definition in Section 14.3),
0w 0w
∆w = (P ) ∆x + (P ) ∆y + e1 ∆x + e2 ∆y,
0x 0 0y 0

where e1, e2 S 0 as ∆x, ∆y S 0. To find dw > dt, we divide this equation through by ∆t
and let ∆t approach zero. The division gives
∆w 0w ∆x 0w ∆y ∆x ∆y
= (P ) + (P ) + e1 + e2 .
∆t 0x 0 ∆t 0y 0 ∆t ∆t ∆t
Letting ∆t approach zero gives
dw ∆w
(t ) = lim
dt 0 ∆t S 0 ∆t

(P ) (t ) + 0 # (t0) + 0 # (t0).
0w dx 0w dy dx dy
= (P ) (t ) +
To remember the Chain Rule, picture the 0x 0 dt 0 0y 0 dt 0 dt dt
diagram below. To find dw>dt, start at w Often we write 0w>0x for the partial derivative 0ƒ>0x, so we can rewrite the Chain
and read down each route to t, multiply-
Rule in Theorem 5 in the form
ing derivatives along the way. Then add
the products.
Chain Rule
dw 0w dx 0w dy
w = f (x, y) Dependent = + .
dt 0x dt 0y dt
variable
'w 'w
'x 'y
However, the meaning of the dependent variable w is different on each side of the preced-
x y Intermediate ing equation. On the left-hand side, it refers to the composite function w = ƒ(x(t), y(t)) as
variables
a function of the single variable t. On the right-hand side, it refers to the function
dx dy w = ƒ(x, y) as a function of the two variables x and y. Moreover, the single derivatives
dt dt
dw>dt, dx>dt, and dy>dt are being evaluated at a point t0 , whereas the partial derivatives
Independent
t 0w>0x and 0w>0y are being evaluated at the point (x0 , y0), with x0 = x(t0) and y0 = y(t0).
variable
dw 'w dx 'w dy With that understanding, we will use both of these forms interchangeably throughout the
= +
dt 'x dt 'y dt text whenever no confusion will arise.
14.4  The Chain Rule 823

The dependency diagram on the preceding page provides a convenient way to remem-
ber the Chain Rule. The “true” independent variable in the composite function is t, whereas
x and y are intermediate variables (controlled by t) and w is the dependent variable.
A more precise notation for the Chain Rule shows where the various derivatives in
Theorem 5 are evaluated:
dw 0ƒ dx 0ƒ dy
(t ) = (x0, y0) (t0) + (x0, y0) (t0),
dt 0 0x dt 0y dt
or, using another notation,
dw 2 0ƒ dx 2 0ƒ dy
= 2 + 2 2 .
dt t0 0x (x , y ) dt t 0y (x , y ) dt t
0 0 0 0 0 0

EXAMPLE 1 Use the Chain Rule to find the derivative of


w = xy
with respect to t along the path x = cos t, y = sin t. What is the derivative’s value at
t = p>2?

Solution We apply the Chain Rule to find dw > dt as follows:


dw 0w dx 0w dy
= +
dt 0x dt 0y dt
0(xy) d 0(xy) d
= (cos t) + (sin t)
0x dt 0y dt
= ( y)(-sin t) + (x)(cos t)
= (sin t)(-sin t) + (cos t)(cos t)
= -sin2 t + cos2 t
= cos 2t.
In this example, we can check the result with a more direct calculation. As a function of t,
1
w = xy = cos t sin t = sin 2t,
2
so
dw d 1 1
= a sin 2tb = (2 cos 2t) = cos 2t.
dt dt 2 2
In either case, at the given value of t,
dw 2 p
= cos a2 b = cos p = -1.
dt t = p>2 2

Functions of Three Variables


You can probably predict the Chain Rule for functions of three intermediate variables, as it
only involves adding the expected third term to the two-variable formula.

THEOREM 6—Chain Rule for Functions of One Independent Variable and


Three Intermediate Variables
If w = ƒ(x, y, z) is differentiable and x, y, and z are differentiable functions of t,
then w is a differentiable function of t and
dw 0w dx 0w dy 0w dz
= + + .
dt 0x dt 0y dt 0z dt
824 Chapter 14 Partial Derivatives

The proof is identical with the proof of Theorem 5 except that there are now three
Here we have three routes from w to
t instead of two, but finding dw>dt is intermediate variables instead of two. The dependency diagram we use for remembering
still the same. Read down each route, the new equation is similar as well, with three routes from w to t.
multiplying derivatives along the way;
then add. EXAMPLE 2 Find dw>dt if
Chain Rule
w = xy + z, x = cos t, y = sin t, z = t.
w = f (x, y, z) Dependent In this example the values of w(t) are changing along the path of a helix (Section 13.1) as t
variable
changes. What is the derivative’s value at t = 0?
'w 'w
'x 'w 'z
'y Solution Using the Chain Rule for three intermediate variables, we have
x y z Intermediate
variables dw 0w dx 0w dy 0w dz
dy = + +
dt 0x dt 0y dt 0z dt
dx dt dz
dt dt = (y)(-sin t) + (x)(cos t) + (1)(1)
Substitute for intermediate
Independent = (sin t)(-sin t) + (cos t)(cos t) + 1 variables.
t variable
dw
=
'w dx
+
'w dy
+
'w dz = -sin2 t + cos2 t + 1 = 1 + cos 2t,
dt 'x dt 'y dt 'z dt so
dw 2
= 1 + cos (0) = 2.
dt t = 0

For a physical interpretation of change along a curve, think of an object whose posi-
tion is changing with time t. If w = T(x, y, z) is the temperature at each point (x, y, z)
along a curve C with parametric equations x = x(t), y = y(t), and z = z(t), then the com-
posite function w = T(x(t), y(t), z(t)) represents the temperature relative to t along the
curve. The derivative dw > dt is then the instantaneous rate of change of temperature due to
the motion along the curve, as calculated in Theorem 6.

Functions Defined on Surfaces


If we are interested in the temperature w = ƒ(x, y, z) at points (x, y, z) on the earth’s sur-
face, we might prefer to think of x, y, and z as functions of the variables r and s that give
the points’ longitudes and latitudes. If x = g(r, s), y = h(r, s), and z = k(r, s), we could
then express the temperature as a function of r and s with the composite function
w = ƒ(g(r, s), h(r, s), k(r, s)).
Under the conditions stated below, w has partial derivatives with respect to both r and s
that can be calculated in the following way.

THEOREM 7—Chain Rule for Two Independent Variables and Three


Intermediate Variables
Suppose that w = ƒ(x, y, z), x = g(r, s), y = h(r, s), and z = k(r, s). If all four
functions are differentiable, then w has partial derivatives with respect to r and s,
given by the formulas
0w 0w 0x 0w 0y 0w 0z
= + +
0r 0x 0r 0y 0r 0z 0r

0w 0w 0x 0w 0y 0w 0z
= + + .
0s 0x 0s 0y 0s 0z 0s

The first of these equations can be derived from the Chain Rule in Theorem 6 by hold-
ing s fixed and treating r as t. The second can be derived in the same way, holding r fixed
and treating s as t. The dependency diagrams for both equations are shown in Figure 14.22.
14.4  The Chain Rule 825

w = f (x, y, z) w = f (x, y, z)
Dependent w
variable 'w 'w
'w 'w
f 'x 'z 'x 'z
'w 'w
'y 'y
Intermediate x y z x y z x y z
variables
'y 'y
g h k 'x 'r 'z 'x 's 'z
'r 'r 's 's
Independent
r, s
variables s
r
w = f( g(r, s), h(r, s), k(r, s)) 'w 'w 'x 'w 'y 'w 'z 'w 'w 'x 'w 'y 'w 'z
= + + = + +
'r 'x 'r dy 'r 'z 'r 's 'x 's 'y 's 'z 's

(a) (b) (c)

FIGURE 14.22 Composite function and dependency diagrams for Theorem 7.

EXAMPLE 3 Express 0w>0r and 0w>0s in terms of r and s if


r
w = x + 2y + z2, x = s, y = r 2 + ln s, z = 2r.

Solution Using the formulas in Theorem 7, we find


0w 0w 0x 0w 0y 0w 0z
= + +
0r 0x 0r 0y 0r 0z 0r

1
= (1)a s b + (2)(2r) + (2z)(2)

1 1
= s + 4r + (4r)(2) = s + 12r Substitute for intermediate variable z.

0w 0w 0x 0w 0y 0w 0z
= + +
0s 0x 0s 0y 0s 0z 0s

r 1 2 r
= (1)a- 2 b + (2)a s b + (2z)(0) = s - 2 .
s s

If ƒ is a function of two intermediate variables instead of three, each equation in


Chain Rule Theorem 7 becomes correspondingly one term shorter.
w = f (x, y)

'w 'w
'x 'y
If w = ƒ(x, y), x = g(r, s), and y = h(r, s), then
x y 0w 0w 0x 0w 0y 0w 0w 0x 0w 0y
= + and = + .
0r 0x 0r 0y 0r 0s 0x 0s 0y 0s
'x 'y
'r 'r

r
'w 'w 'x 'w 'y Figure 14.23 shows the dependency diagram for the first of these equations. The dia-
= +
'r 'x 'r 'y 'r gram for the second equation is similar; just replace r with s.
FIGURE 14.23 Dependency diagram for
the equation
EXAMPLE 4 Express 0w>0r and 0w>0s in terms of r and s if
0w 0w 0x 0w 0y
= + .
0r 0x 0r 0y 0r w = x2 + y 2, x = r - s, y = r + s.
826 Chapter 14 Partial Derivatives

Solution The preceding discussion gives the following.


0w 0w 0x 0w 0y 0w 0w 0x 0w 0y
= + = +
0r 0x 0r 0y 0r 0s 0x 0s 0y 0s
Substitute for
= (2x)(1) + (2y)(1) = (2x)(-1) + (2y)(1) the intermediate
variables.
= 2(r - s) + 2(r + s) = -2(r - s) + 2(r + s)
= 4r = 4s

If ƒ is a function of a single intermediate variable x, our equations are even simpler.

If w = ƒ(x) and x = g(r, s), then

0w dw 0x 0w dw 0x
= and = .
0r dx 0r 0s dx 0s

In this case, we use the ordinary (single-variable) derivative, dw > dx. The dependency
Chain Rule
diagram is shown in Figure 14.24.

w = f (x)

dw
dx Implicit Differentiation Revisited
x The two-variable Chain Rule in Theorem 5 leads to a formula that takes some of the alge-
'x 'x bra out of implicit differentiation. Suppose that
'r 's
1. The function F(x, y) is differentiable and
2. The equation F(x, y) = 0 defines y implicitly as a differentiable function of x, say
r s
y = h(x).
'w dw 'x
=
'r dx 'r Since w = F(x, y) = 0, the derivative dw > dx must be zero. Computing the derivative
'w dw 'x from the Chain Rule (dependency diagram in Figure 14.25), we find
=
's dx 's

FIGURE 14.24 Dependency diagram for dw dx dy Theorem 5 with


0 = = Fx + Fy t = x and ƒ = F
differentiating ƒ as a composite function of dx dx dx
r and s with one intermediate variable.
= Fx # 1 + Fy #
dy
.
dx
w = F(x, y)

'w = F If Fy = 0w>0y ≠ 0, we can solve this equation for dy > dx to get


x Fy = 'w
'x 'y
dy Fx
x y = h(x) = - .
dx Fy

dx dy We state this result formally.


=1 = h′(x)
dx dx

x
dw dy
= Fx • 1 + Fy •
dx dx THEOREM 8—A Formula for Implicit Differentiation
Suppose that F(x, y) is differentiable and that the equation F(x, y) = 0 defines y
FIGURE 14.25 Dependency diagram
as a differentiable function of x. Then at any point where Fy ≠ 0,
for differentiating w = F(x, y) with
respect to x. Setting dw>dx = 0 leads to a dy Fx
simple computational formula for implicit = - . (1)
dx Fy
differentiation (Theorem 8).
14.4  The Chain Rule 827

EXAMPLE 5 Use Theorem 8 to find dy > dx if y 2 - x2 - sin xy = 0.

Solution Take F(x, y) = y 2 - x2 - sin xy. Then

dy Fx -2x - y cos xy 2x + y cos xy


= - = - = .
dx Fy 2y - x cos xy 2y - x cos xy

This calculation is significantly shorter than a single-variable calculation using implicit


differentiation.

The result in Theorem 8 is easily extended to three variables. Suppose that the equa-
tion F(x, y, z) = 0 defines the variable z implicitly as a function z = ƒ(x, y). Then for all
(x, y) in the domain of ƒ, we have F(x, y, ƒ(x, y)) = 0. Assuming that F and ƒ are differen-
tiable functions, we can use the Chain Rule to differentiate the equation F(x, y, z) = 0
with respect to the independent variable x:

0F 0x 0F 0y 0F 0z
0 = + +
0x 0x 0y 0x 0z 0x

= Fx # 1 + Fy # 0 + Fz #
y is constant when
0z
, differentiating with
0x respect to x.

so

0z
Fx + Fz = 0.
0x

A similar calculation for differentiating with respect to the independent variable y gives

0z
Fy + Fz = 0.
0y

Whenever Fz ≠ 0, we can solve these last two equations for the partial derivatives of
z = ƒ(x, y) to obtain

0z Fx 0z Fy
= - and = - . (2)
0x Fz 0y Fz

An important result from advanced calculus, called the Implicit Function Theorem,
states the conditions for which our results in Equations (2) are valid. If the partial deriva-
tives Fx, Fy, and Fz are continuous throughout an open region R in space containing the
point (x0, y0, z0), and if for some constant c, F(x0, y0, z0) = c and Fz(x0, y0, z0) ≠ 0, then
the equation F(x, y, z) = c defines z implicitly as a differentiable function of x and y near
(x0, y0, z0), and the partial derivatives of z are given by Equations (2).

0z 0z
EXAMPLE 6 Find and at (0, 0, 0) if x3 + z2 + yexz + z cos y = 0.
0x 0y
Solution Let F(x, y, z) = x3 + z2 + yexz + z cos y. Then
Fx = 3x2 + zyexz, Fy = exz - z sin y, and Fz = 2z + xyexz + cos y.
828 Chapter 14 Partial Derivatives

Since F(0, 0, 0) = 0, Fz(0, 0, 0) = 1 ≠ 0, and all first partial derivatives are continuous,
the Implicit Function Theorem says that F(x, y, z) = 0 defines z as a differentiable func-
tion of x and y near the point (0, 0, 0). From Equations (2),
0z Fx 3x2 + zyexz 0z Fy exz - z sin y
= - = - and = - = - .
0x Fz 2z + xyexz + cos y 0y Fz 2z + xyexz + cos y
At (0, 0, 0) we find
0z 0 0z 1
= - = 0 and = - = -1.
0x 1 0y 1

Functions of Many Variables


We have seen several different forms of the Chain Rule in this section, but each one is just
a special case of one general formula. When solving particular problems, it may help to
draw the appropriate dependency diagram by placing the dependent variable on top, the
intermediate variables in the middle, and the selected independent variable at the bottom.
To find the derivative of the dependent variable with respect to the selected independent
variable, start at the dependent variable and read down each route of the dependency dia-
gram to the independent variable, calculating and multiplying the derivatives along each
route. Then add the products found for the different routes.
In general, suppose that w = ƒ(x, y, . . . , y) is a differentiable function of the inter-
mediate variables x, y, . . . , y (a finite set) and the x, y, . . . , y are differentiable functions
of the independent variables p, q, . . . , t (another finite set). Then w is a differentiable
function of the variables p through t, and the partial derivatives of w with respect to these
variables are given by equations of the form
0w 0w 0x 0w 0y 0w 0y
= + + g + .
0p 0x 0p 0y 0p 0y 0p

The other equations are obtained by replacing p by q, . . . , t, one at a time.


One way to remember this equation is to think of the right-hand side as the dot prod-
uct of two vectors with components
0w 0w 0w 0x 0y 0y
a , ,..., b and a , , . . . , b.
0x 0y 0y 0p 0p 0p
(++++)++++* (++++)++++*
Derivatives of w with Derivatives of the intermediate
respect to the variables with respect to the
intermediate variables selected independent variable

EXERCISES 14.4
Chain Rule: One Independent Variable 5. w = 2yex - ln z, x = ln ( t 2 + 1 ) , y = tan-1 t, z = et ;
In Exercises 1–6, (a) express dw > dt as a function of t, both by using t = 1
the Chain Rule and by expressing w in terms of t and differentiating
6. w = z - sin xy, x = t, y = ln t, z = et - 1 ; t = 1
directly with respect to t. Then (b) evaluate dw > dt at the given value
of t. Chain Rule: Two and Three Independent Variables
1. w = x + y , x = cos t, y = sin t; t = p
2 2 In Exercises 7 and 8, (a) express 0z>0u and 0z>0y as functions of u
and y both by using the Chain Rule and by expressing z directly in
2. w = x + y , x = cos t + sin t, y = cos t - sin t; t = 0
2 2
terms of u and y before differentiating. Then (b) evaluate 0z>0u and
x y 0z>0y at the given point (u, y).
3. w = z + z , x = cos2 t, y = sin2 t, z = 1>t ; t = 3
7. z = 4ex ln y, x = ln (u cos y), y = u sin y;
4. w = ln ( x2 + y 2 + z2 ) , x = cos t, y = sin t, z = 4 2t ; (u, y) = (2, p>4)
t = 3
14.4  The Chain Rule 829

8. z = tan-1 (x>y), x = u cos y, y = u sin y; 27. x2 + xy + y 2 - 7 = 0, (1, 2)


(u, y) = (1.3, p>6) 28. xey + sin xy + y - ln 2 = 0, (0, ln 2)
In Exercises 9 and 10, (a) express 0w>0u and 0w>0y as functions of u 29. (x3 - y 4)6 + ln (x2 + y) = 1, (- 1, 0)
and y both by using the Chain Rule and by expressing w directly in 30. xex y - yex = x - y + 2,
2
(1, 1)
terms of u and y before differentiating. Then (b) evaluate 0w>0u and
0w>0y at the given point (u, y). Find the values of 0z>0x and 0z>0y at the points in Exercises 31–34.
9. w = xy + yz + xz, x = u + y, y = u - y, z = uy; 31. z3 - xy + yz + y 3 - 2 = 0, (1, 1, 1)
(u, y) = (1>2, 1) 1 1 1
32. x + y + z - 1 = 0, (2, 3, 6)
10. w = ln ( x2 + y 2 + z2 ) , x = uey sin u, y = uey cos u,
z = uey; (u, y) = (-2, 0) 33. sin (x + y) + sin ( y + z) + sin (x + z) = 0, (p, p, p)
In Exercises 11 and 12, (a) express 0u>0x, 0u>0y, and 0u>0z as func- 34. xey + yez + 2 ln x - 2 - 3 ln 2 = 0, (1, ln 2, ln 3)
tions of x, y, and z both by using the Chain Rule and by expressing u
directly in terms of x, y, and z before differentiating. Then (b) evaluate Finding Partial Derivatives at Specified Points
0u>0x, 0u>0y, and 0u>0z at the given point (x, y, z). 35. Find 0w>0r when r = 1, s = - 1 if w = (x + y + z)2,
x = r - s, y = cos (r + s), z = sin (r + s).
p - q
11. u = q - r , p = x + y + z, q = x - y + z, 36. Find 0w>0y when u = - 1, y = 2 if w = xy + ln z,
x = y2 >u, y = u + y, z = cos u.
r = x + y - z; (x, y, z) = 1 23, 2, 1 2
-1 37. Find 0w>0y when u = 0, y = 0 if w = x2 + ( y>x),
12. u = e sin
qr
p, p = sin x, q = z2 ln y, r = 1>z;
x = u - 2y + 1, y = 2u + y - 2.
(x, y, z) = (p>4, 1>2, -1>2)
38. Find 0z>0u when u = 0, y = 1 if z = sin xy + x sin y,
Using a Dependency Diagram x = u2 + y2, y = uy.
In Exercises 13–24, draw a dependency diagram and write a Chain
39. Find 0z>0u and 0z>0y when u = ln 2, y = 1 if z = 5 tan-1 x and
Rule formula for each derivative.
x = eu + ln y.
dz 40. Find 0z>0u and 0z>0y when u = 1, y = -2 if z = ln q and
13. for z = ƒ(x, y), x = g(t), y = h(t)
dt q = 1y + 3 tan-1 u.
dz
14. for z = ƒ(u, y, w), u = g(t), y = h(t), w = k(t) Theory and Examples
dt
0w 0w 0w 0w
15. and for w = h(x, y, z), x = ƒ(u, y), y = g(u, y), 41. Assume that w = ƒ ( s3 + t 2 ) and ƒ′(x) = ex. Find and .
0u 0y 0t 0s
z = k(u, y)
s 0ƒ 0ƒ x2
0w 0w 42. Assume that w = ƒats2, t b, (x, y) = xy, and (x, y) = .
16. and for w = ƒ(r, s, t), r = g(x, y), s = h(x, y), 0x 0y 2
0x 0y
t = k(x, y) 0w 0w
Find and .
0t 0s
0w 0w
17. and for w = g(x, y), x = h(u, y), y = k(u, y)
0u 0y 43. Assume that z = ƒ(x, y), x = g(t), y = h(t), ƒx(2, -1) = 3,
0w 0w and ƒy(2, - 1) = - 2. If g(0) = 2, h(0) = - 1, g′(0) = 5, and
18. and for w = g(u, y), u = h(x, y), y = k(x, y)
0x 0y dz 2
0z 0z h′(0) = - 4, find .
dt t = 0
19. and for z = ƒ(x, y), x = g(t, s), y = h(t, s)
0t 0s
44. Assume that z = ƒ(x, y)2, x = g(t), y = h(t), ƒx(1, 0) = - 1,
0y ƒy(1, 0) = 1, and ƒ(1, 0) = 2. If g(3) = 1, h(3) = 0, g′(3) = -3,
20. for y = ƒ(u), u = g(r, s)
0r
dz 2
0w 0w and h′(3) = 4, find .
21. and for w = g(u), u = h(s, t) dt t = 3
0s 0t
0w 45. Assume that z = ƒ(w), w = g(x, y), x = 2r 3 - s2, and y = res .
22. for w = ƒ(x, y, z, y), x = g( p, q), y = h( p, q),
0p If gx(2, 1) = - 3, gy(2, 1) = 2, ƒ′(7) = - 1, and g(2, 1) = 7,
z = j(p, q), y = k(p, q)
0z 2 0z
0w 0w find and 2 .
23. and for w = ƒ(x, y), x = g(r), y = h(s) 0r r = 1, s = 0 0s r = 1, s = 0
0r 0s
46. Assume that z = ln (ƒ(w)), w = g(x, y), x = 2r - s, and
0w y = r 2s. If gx(2, - 9) = - 1, gy(2, - 9) = 3, ƒ′(- 2) = 2,
24. for w = g(x, y), x = h(r, s, t), y = k(r, s, t)
0s
0z 2
Implicit Differentiation ƒ(-2) = 5, and g(2, -9) = - 2, find and
0r r = 3, s = - 1
Assuming that the equations in Exercises 25–30 define y as a differen-
tiable function of x, use Theorem 8 to find the value of dy > dx at the 0z 2
.
given point. 0s r = 3, s = - 1
25. x3 - 2y 2 + xy = 0, (1, 1)
26. xy + y 2 - 3x - 3 = 0, (- 1, 1)
830 Chapter 14 Partial Derivatives

47. Changing voltage in a circuit The voltage V in a circuit that 54. A space curve Let w = x2e2y cos 3z. Find the value of dw > dt
satisfies the law V = IR is slowly dropping as the battery wears at the point (1, ln 2, 0) on the curve x = cos t, y = ln (t + 2),
out. At the same time, the resistance R is increasing as the resistor z = t.
heats up. Use the equation
55. Temperature on a circle Let T = ƒ(x, y) be the temperature at
dV 0V dI 0V dR the point (x, y) on the circle x = cos t, y = sin t, 0 … t … 2p and
= +
dt 0I dt 0R dt suppose that
to find how the current is changing at the instant when R = 0T 0T
600 ohms, I = 0.04 amp, dR>dt = 0.5 ohm>sec, and dV>dt = = 8x - 4y, = 8y - 4x.
0x 0y
-0.01 volt>sec.
a. Find where the maximum and minimum temperatures on the
V
+ − circle occur by examining the derivatives dT > dt and d 2T>dt 2.
Battery b. Suppose that T = 4x2 - 4xy + 4y 2. Find the maximum and
minimum values of T on the circle.
I 56. Temperature on an ellipse Let T = g(x, y) be the temperature
at the point (x, y) on the ellipse
x = 2 22 cos t, y = 22 sin t, 0 … t … 2p,
R
and suppose that
48. Changing dimensions in a box The lengths a, b, and c of the
edges of a rectangular box are changing with time. At the instant 0T 0T
= y, = x.
in question, a = 1 m, b = 2 m, c = 3 m, da>dt = db>dt = 0x 0y
1 m>sec, and dc>dt = - 3 m>sec. At what rates are the box’s vol- a. Locate the maximum and minimum temperatures on the
ume V and surface area S changing at that instant? Are the box’s ellipse by examining dT > dt and d 2T>dt 2.
interior diagonals increasing in length or decreasing?
b. Suppose that T = xy - 2. Find the maximum and minimum
49. If ƒ(u, y, w) is differentiable and u = x - y, y = y - z, and values of T on the ellipse.
w = z - x, show that
57. The temperature T = T(x, y) in °C at point (x, y)
0ƒ 0ƒ 0ƒ satisfies Tx(1, 2) = 3 and Ty(1, 2) = - 1. If x = e2t - 2 cm and
+ + = 0.
0x 0y 0z y = 2 + ln t cm, find the rate at which the temperature T
50. Polar coordinates Suppose that we substitute polar coordi- changes when t = 1 sec.
nates x = r cos u and y = r sin u in a differentiable function 58. A bug crawls on the surface z = x2 - y 2 directly above
w = ƒ(x, y). a path in the xy-plane given by x = ƒ(t) and y = g(t). If
a. Show that ƒ(2) = 4, ƒ′(2) = - 1, g(2) = - 2, and g′(2) = - 3, then at what
rate is the bug’s elevation z changing when t = 2?
0w
= ƒx cos u + ƒy sin u Differentiating Integrals Under mild continuity restrictions, it is
0r
true that if
and b

1 0w F(x) = g(t, x) dt,


r 0u = - ƒx sin u + ƒy cos u. La
b
b. Solve the equations in part (a) to express ƒx and ƒy in terms then F′(x) = gx(t, x) dt. Using this fact and the Chain Rule, we can
of 0w>0r and 0w>0u. La
c. Show that find the derivative of
ƒ(x)
0w 2 1 0w 2 F(x) = g(t, x) dt
(ƒx) + (ƒy) = a b + 2 a b .
2 2
0r r 0u La
51. Laplace equations Show that if w = ƒ(u, y) satisfies the La- by letting
place equation ƒuu + ƒyy = 0 and if u = ( x2 - y 2 ) >2 and u
y = xy, then w satisfies the Laplace equation wxx + wyy = 0. G(u, x) = g(t, x) dt,
La
52. Laplace equations Let w = ƒ(u) + g(y), where u = x + iy,
y = x - iy, and i = 2- 1. Show that w satisfies the Laplace where u = ƒ(x). Find the derivatives of the functions in Exercises 59
equation wxx + wyy = 0 if all the necessary functions are differ- and 60.
x2
entiable.
59. F(x) = 2t 4 + x3 dt
53. Extreme values on a helix Suppose that the partial derivatives L0
of a function ƒ(x, y, z) at points on the helix x = cos t, y = sin t, 1
z = t are 60. F(x) = 2t 3 + x2 dt
ƒx = cos t, ƒy = sin t, ƒz = t + t - 2.
2 Lx2

At what points on the curve, if any, can ƒ take on extreme values?