13.

9 Lagrange Multipliers 1127

a. Find an equation of the least-squares line for these data. 63. Let f(x, y)  3x 2  6x  4y 2  4y  3.
b. Use the result of (a) to estimate the life expectancy at 65 a. Show that f has no minimum value.
of a male in 2040. How does this result compare with the cas b. Find the maximum and minimum values of f in the
given data for that year? region D  {(x, y) 冟 x 2  y 2 1}.
c. Use the result of (a) to estimate the life expectancy at 65 Hint: On the boundary of D, let x  cos t, y  sin t for
of a male in 2030. 0 t 2p.
Source: U.S. Census Bureau.
64. Let f(x, y)  Ax 2  2Bxy  Cy 2, where B 2  4AC  0.
61. Operations Management Consulting Spending Refer to Exer- Find conditions in terms of A, B, and C such that f has a
cise 57. The following table gives the projected operations relative minimum at (0, 0); a relative maximum at (0, 0);
management consulting spending (in billions of dollars) and a saddle point at (0, 0).
from 2005 through 2010. Here, x  5 corresponds to 2005.
In Exercises 65–68, determine whether the statement is true or
Year, x 5 6 7 8 9 10 false. If it is true, explain why it is true. If it is false, give an
example to show why it is false.
Spending, y 40 43.2 47.4 50.5 53.7 56.8
65. If f(x, y) has a relative maximum at (a, b), then fx (a, b)  0
and fy(a, b)  0.
a. Find an equation of the least-squares line for these data.
b. Use the results of part (a) to estimate the average rate of 66. Let h(x, y)  f(x)  t(y), where f and t have second-order
change of operations management consulting spending derivatives. If the graph of f is concave upward on (⬁, ⬁)
from 2005 through 2010. and the graph of t is concave downward on (⬁, ⬁), then h
c. Use the results of part (a) to estimate the amount of cannot have a relative maximum or a relative minimum at
spending on operations management consulting in 2011, any point.
assuming that the trend continues. 67. If §f(a, b)  0 , then f has a relative extremum at (a, b).
Source: Kennedy Information. 68. If f(x, y) has continuous second-order partial derivatives and
62. Let f(x, y)  x 2  y 2  2xy  2. fxx(x, y)  fyy(x, y)  0 and fxy (x, y)  0 for all (x, y), then f
a. Show that f has no maximum or minimum values. cannot have a relative extremum.
cas b. Find the maximum and minimum values of f in the
region D  {(x, y) 冟 x 2  4y 2 4}.
Hint: On the boundary of D, let x  2 cos t, y  sin t for
0 t 2p.

13.9 Lagrange Multipliers

Constrained Maxima and Minima
y Many practical optimization problems involve maximizing or minimizing an objective
function subject to one or more constraints, or side conditions. In Example 5 of Sec-
tion 13.8 we discussed the problem of maximizing the (volume) function
V  f(x, y, z)  xyz
a y
subject to the constraint
x t(x, y, z)  x  2y  2z  108
In this case the constraint expresses the condition that the combined length plus girth
of a package is 108 in. (the maximum allowed by postal regulations).
As another example, consider a problem encountered in the construction of an AC
transformer. Here, we are required to find the cross-shaped iron core of largest surface
FIGURE 1 area that can be inserted into a coil of radius a (Figure 1). In terms of x and y we see
We want to find the core of largest that the surface area of the iron core is
surface area that can be inserted into
a coil of radius a. S  4xy  4y(x  y)  8xy  4y 2
1128 Chapter 13 Functions of Several Variables

Next, observe that x and y must satisfy the equation x 2  y 2  a 2. Therefore, the prob-
lem is equivalent to one of maximizing the objective function
f(x, y)  8xy  4y 2
subject to the constraint
t(x, y)  x 2  y 2  a 2
We will complete the solution of this problem in Example 2.
Figure 2a shows the graph of a function f defined by the equation z  f(x, y).
Observe that f has an absolute minimum at (0, 0) and an absolute minimum value of
0. However, if the independent variables x and y are subjected to a constraint of the
form t(x, y)  k, then the points (x, y, z) that satisfy both z  f(x, y) and t(x, y)  k
lie on the curve C, the intersection of the surface z  f(x, y) and the cylinder t(x, y)  k
(Figure 2b). From the figure you can see that the absolute minimum of f subject to the
constraint t(x, y)  k occurs at the point (a, b). Furthermore, f has the constrained
absolute minimum value f(a, b) rather than the unconstrained absolute minimum value
of 0 at (0, 0).
z z

z  f (x, y) z  f (x, y)

(a, b, f(a, b))

FIGURE 2 (0, 0)
The function f has an unconstrained
minimum value of 0, but it has a y g(x, y)  k (a, b) y
constrained minimum value of f(a, b) x x
when subjected to the constraint
t(x, y)  k. (a) f is not subject to any constraints. (b) f is subject to a constraint.

The problem that we discussed at the beginning of this section (maximizing the
volume of a box subject to a given constraint) was first solved in Section 13.8. Recall
the method of solution that we used:
First, we solved the constraint equation
t(x, y, z)  x  2y  2z  108
for x in terms of y and z. We then substituted this expression for x into the equation
V  f(x, y, z)  xyz
thereby obtaining an expression for V involving the variables y and z and satisfying the
constraint equation. Next, we found the maximum of V by treating V as an uncon-
strained function of y and z.
The major drawback of this method is that it relies on our ability to solve the con-
straint equation t(x, y)  k for one variable explicitly in terms of the other (or
t(x, y, z)  k for one variable explicitly in terms of the other two variables in the case
of a constraint involving three variables). This might not always be possible or con-
venient. Moreover, even when we are able to solve the constraint equation t(x, y)  k
for y explicitly in terms of x, the resulting function of one variable that is obtained by
substituting this expression for y into the objective function f(x, y) might turn out to
be unnecessarily complicated.
 13.9 Lagrange Multipliers 1129

The Method of Lagrange Multipliers

We will now consider a method, called the method of Lagrange multipliers (named
after the French mathematician Joseph Lagrange, 1736–1813), which obviates the need
to solve the constraint equation for one variable in terms of the other variables. To see
how this method works, let’s reexamine the problem of finding the absolute minimum
of the objective function f subject to the constraint t(x, y)  k that we considered ear-
lier. Figure 3a shows the level curves of f drawn in the xyz-coordinate system. These
level curves are reproduced in the xy-plane in Figure 3b.

z
y
z  f (x, y)
g(x, y)  k
z  c1
z  c2 (a, b)

C z  f (a, b)

zc (a, b, f (a, b)) x

z  c2 1
z  f (a, b) z  c3
zc
z  c4 3
z  c4
g(x, y)  k (a, b) y

FIGURE 3 (a) The level curves of f in the xyz-plane (b) The level curves of f in the xy-plane

Observe that the level curves of f with equations f(x, y)  c, where c  f(a, b),
have no points in common with the graph of the constraint equation t(x, y)  k (for
example, the level curves f(x, y)  c1 and f(x, y)  c2 shown in Figure 3). Thus, points
lying on these curves are not candidates for the constrained minimum of f.
On the other hand, the level curves of f with equation f(x, y)  c, where
c  f(a, b), do intersect the graph of the constraint equation t(x, y)  k (such as the
level curves of f(x, y)  c3 and f(x, y)  c4). These points of intersection are candi-
dates for the constrained minimum of f.
Finally, observe that the larger c is for c  f(a, b), the larger the value f(x, y) is for
(x, y) lying on the level curve t(x, y)  k. This observation suggests that we can find
the constrained minimum of f by choosing the smallest value of c so that the level curve
f(x, y)  c still intersects the curve t(x, y)  k. At such a point (a, b) the level curve
of f just touches the graph of the constraint equation t(x, y)  k. That is, the two curves
have a common tangent at (a, b) (see Figure 3b). Equivalently, their normal lines at
this point coincide. Putting it yet another way, the gradient vectors §f(a, b) and §t(a, b)
have the same direction, so §f(a, b)  l§t(a, b) for some scalar l (lambda).
A similar result holds for the problem of maximizing or minimizing a function f of
three variables defined by w  f(x, y, z) and subject to the constraint t(x, y, z)  k.
In this situation, f has a constrained maximum or constrained minimum at a point
(a, b, c) where the level surface f(x, y, z)  f(a, b, c) is tangent to the level surface
t(x, y, z)  k. But this means that the normals of these surfaces, and therefore their
gradient vectors, at the point (a, b, c) must be parallel to each other. Thus, there is a
scalar l such that §f(a, b, c)  l§t(a, b, c).
These geometric arguments suggest the following theorem.
1130 Chapter 13 Functions of Several Variables

THEOREM 1 Lagrange’s Theorem

Let f and t have continuous first partial derivatives in some region D in the plane.
If f has an extremum at a point (a, b) on the smooth constraint curve t(x, y)  c
lying in D and §t(a, b)  0, then there is a real number l such that
§f(a, b)  l§t(a, b)

The number l in Theorem 1 is called a Lagrange multiplier.

PROOF Suppose that the smooth curve C described by t(x, y)  c is represented by

the vector function
r(t)  x(t)i  y(t)j, rⴕ(t)  0
where x¿ and y¿ are continuous on an open interval I (Figure 4). Then the values assumed
by f on C are given by
h(t)  f(x(t), y(t))
for t in I. Suppose that f has an extreme value at (a, b). If t 0 is the point in I correspond-
ing to the point (a, b) , then h has an extreme value at t 0. Therefore, h¿(t 0)  0. Using
the Chain Rule, we have
h¿(t 0)  fx (x(t 0), y(t 0))x¿(t 0)  fy(x(t 0), y(t 0))y¿(t 0)
 fx (a, b)x¿(t 0)  fy (a, b)y¿(t 0)
 §f(a, b) ⴢ rⴕ(t 0)  0
This shows that §f(a, b) is orthogonal to rⴕ(t 0) . But as we demonstrated in Sec-
tion 13.7, §t(a, b) is orthogonal to rⴕ(t 0). Therefore, the gradient vectors §f(a, b) and
§t(a, b) are parallel, so there is a number l such that §f(a, b)  l §t(a, b).

y
r(t0)
◊ g(a, b)

(a, b)

r(t) C g(x, y)  c

t0 t 0 x

(a) The parameter interval I (b) The smooth curve C is represented by the
FIGURE 4 vector function r(t).

The proof of Lagrange’s Theorem for functions of three variables is similar to that
for functions of two variables. In the case involving three variables, level surfaces rather
than level curves are involved. Lagrange’s Theorem leads to the following procedure
for finding the constrained extremum values of functions. We state it for the case of
functions of three variables.
 13.9 Lagrange Multipliers 1131

The Method of Lagrange Multipliers

Suppose f and t have continuous first partial derivatives. To find the maximum
and minimum values of f subject to the constraint t(x, y, z)  k (assuming that
these extreme values exist and that §t  0 on t(x, y, z)  k):
1. Solve the equations
§f(x, y, z)  l§t(x, y, z) and t(x, y, z)  k
for x, y, z, and l.
2. Evaluate f at each solution point found in Step 1. The largest value yields
the constrained maximum of f, and the smallest value yields the con-
strained minimum of f.

Note Since
§f(x, y, z)  fx (x, y, z)i  fy (x, y, z)j  fz (x, y, z)k
and
§t(x, y, z)  tx (x, y, z)i  ty (x, y, z)j  tz (x, y, z)k
we see, by equating like components, that the vector equation
§f(x, y, z)  l §t(x, y, z)
is equivalent to the three scalar equations
fx (x, y, z)  ltx(x, y, z) , fy (x, y, z)  lty(x, y, z) , and fz(x, y, z)  ltz(x, y, z)
These scalar equations together with the constraint equation t(x, y, z)  k give a sys-
tem of four equations to be solved for the four unknowns x, y, z, and l.

EXAMPLE 1 Find the maximum and minimum values of the function f(x, y) 
x 2  2y subject to x 2  y 2  9.

Solution The constraint equation is t(x, y)  x 2  y 2  9. Since

§f(x, y)  2x i  2j and §t(x, y)  2x i  2yj
the equation §f(x, y)  l§t(x, y) becomes
2xi  2j  l(2x i  2yj)  2lx i  2ly j
Equating like components and rewriting the constraint equation lead to the follow-
ing system of three equations in the three variables x, y, and l:
2x  2lx (1a)
2  2ly (1b)
x y 9
2 2
(1c)
From Equation (1a) we have
2x(1  l)  0
1132 Chapter 13 Functions of Several Variables

so x  0, or l  1. If x  0, then Equation (1c) gives y  3. If l  1, then

Equation (1b) gives y  1, which upon substitution into Equation (1c) yields
x  212. Therefore, f has possible extreme values at the points (0, 3), (0, 3),
(212, 1), and (2 12, 1). Evaluating f at each of these points gives
f(0, 3)  6, f(0, 3)  6, f(2 12, 1)  10, and f(2 12, 1)  10
We conclude that the maximum value of f on the circle x  y  9 is 10, attained at
2 2

the points (212, 1) and (2 12, 1), and that the minimum value of f on the cir-
cle is 6, attained at the point (0, 3) .
Figure 5 shows the graph of the constraint equation x 2  y 2  9 and some level
curves of the objective function f. Observe that the extreme values of f are attained at
the points where the level curves of f are tangent to the graph of the constraint equation.

f (x, y)  x2  2y  6 y f (x, y)  x2  2y  0

(0, 3)

f (x, y)  x2  2y  10

0 x
(2 √ 2, 1) (2 √ 2, 1)
FIGURE 5
The extreme values of f occur at
the points where the level curves x2  y2  9
of f are tangent to the graph of
the constraint equation (the circle).

y EXAMPLE 2 Complete the solution to the problem posed at the beginning of this
section: Find the cross-shaped iron core of largest surface area that can be inserted into
a coil of radius a (Figure 6).

a y Solution Recall that the problem reduces to one of finding the largest value of the ob-
jective function f(x, y)  8xy  4y 2 subject to the constraint t(x, y)  x 2  y 2  a 2.
Since
x
§f(x, y)  8y i  (8x  8y)j and §t(x, y)  2xi  2y j
the equation §f(x, y)  l§t(x, y) becomes
8y i  (8x  8y)j  l(2x i  2yj)  2lxi  2ly j
FIGURE 6 Equating like components and rewriting the constraint equation, we get the following
A cross-shaped iron core of largest system of three equations in the three variables x, y, and l:
surface area is to be inserted into the
coil. 8y  2lx (2a)
8x  8y  2ly (2b)
x y a
2 2 2
(2c)
Solving Equation (2a) for y, we obtain y  14 lx. Substituting this expression for y into
Equation (2b) gives
1 2
8x  2lx  lx
2
 13.9 Lagrange Multipliers 1133

or
x(l2  4l  16)  0
Observe that x  0; otherwise, Equation (2a) implies that y  0, so Equation (2c)
becomes 0  a 2, which is impossible. So we have l2  4l  16  0. Using the quad-
ratic formula, we obtain
4 116  64
l  2 215
2
Observe that l must be positive; otherwise, Equation (2a) implies that x or y must be
negative. So we choose l  2  215 ⬇ 2.4721. Next, substituting y  14 lx into
Equation (2c) gives
1 2 2
x2  l x  a2
16
l2
x 2 a1  b  a2
16
l2  16
x2a b  a2
16
or
4a 4
y x ⬇ a Recall that l ⬇ 2.4721.
2l  16
2
2(2.4721) 2  16
⬇ 0.8507a
(0.8507a, 0.5258a) Finally,
1 1
a a x y lx ⬇ (2.4721)(0.8507a) ⬇ 0.5258a
4 4
Therefore, the core will have the largest surface area if x ⬇ 0.8507a and y ⬇ 0.5258a,
where a is the radius of the coil.
FIGURE 7 Figure 7 shows the graph of the constraint equation x 2  y 2  a 2 (the circle of
The maximum value of f occurs at the radius a centered at the origin) and several level curves of the objective function f.
point where the level curve of f is Once again, observe that the maximum value of f, f(0.8507a, 0.5258a) ⬇ 2.4725a 2,
tangent to the level curve of the occurs at the point (0.8507a, 0.5258a) , where the level curve of f is tangent to the graph
constraint equation. of the constraint equation.

EXAMPLE 3 Find the dimensions of a rectangular package having the greatest pos-
sible volume and satisfying the postal regulation that specifies that the combined length
and girth of the package may not exceed 108 inches. (See Example 5 in Section 13.8.)

Solution Recall that to solve this problem, we need to find the largest value of the vol-
ume function f(x, y, z)  xyz subject to the constraint t(x, y, z)  x  2y  2z  108.
To solve this problem using the method of Lagrange multipliers, observe that
§f(x, y, z)  yz i  xz j  xy k and §t(x, y, z)  i  2j  2k
so the equation §f(x, y, z)  l§t(x, y, z) becomes
yz i  xz j  xy k  l(i  2j  2k)
1134 Chapter 13 Functions of Several Variables

Equating components and rewriting the constraint equation give the following system
of four equations in the four variables x, y, z, and l:
yz  l (3a)
xz  2l (3b)
xy  2l (3c)
x  2y  2z  108 (3d)
Substituting Equation (3a) into Equation (3b) yields
xz  2yz or z(x  2y)  0
Since z  0, we have x  2y. Next, substituting Equation (3a) into Equation (3c) gives
xy  2yz or y(x  2z)  0
Since y  0, we have x  2z. Equating the two expressions for x just obtained gives
2y  2z or yz
Finally, substituting the expressions for x and y into Equation (3d) gives
2z  2z  2z  108 or z  18
So y  18 and x  2(18)  36. Therefore, the dimensions of the package are
18 in. 18 in. 36 in., as was obtained before.

Interpreting Our Results

Geometrically, this problem is one of finding the point on the plane x  2y  2z  108
at which f(x, y, z)  xyz has the largest value. The point (36, 18, 18) is precisely the
point at which the level surface xyz  f(36, 18, 18)  11,664 is tangent to the plane
x  2y  2z  108.

EXAMPLE 4 Find the dimensions of the open rectangular box of maximum volume
that can be constructed from a rectangular piece of cardboard having an area of 48 ft2.
What is the volume of the box?

Solution Let the length, width, and height of the box (in feet) be x, y, and z, as shown
in Figure 8. Then the volume of the box is V  xyz. The area of the bottom of the box
plus the area of the four sides is
z
xy  2xz  2yz
x square feet, and this is equal to the area of the cardboard; that is,
y
xy  2xz  2yz  48
FIGURE 8
An open rectangular box of maximum Thus, the problem is one of maximizing the objective function
volume is to be constructed from a
f(x, y, z)  xyz
piece of cardboard. What are the
dimensions of the box? subject to the constraint
t(x, y, z)  xy  2xz  2yz  48
Since
§f(x, y, z)  yz i  xz j  xy k
 13.9 Lagrange Multipliers 1135

and
§t(x, y, z)  (y  2z)i  (x  2z)j  (2x  2y)k
the equation §f(x, y, z)  l §t(x, y, z) becomes
yz i  xz j  xy k  l[(y  2z)i  (x  2z)j  (2x  2y)k]
Equating like components and rewriting the constraint equation give the following sys-
tem of four equations in the unknowns x, y, z, and l:
yz  l(y  2z) (4a)
xz  l(x  2z) (4b)
xy  l(2x  2y) (4c)
xy  2xz  2yz  48 (4d)
Multiplying Equations (4a), (4b), and (4c) by x, y, and z, respectively, gives
xyz  l(xy  2xz) (5a)
xyz  l(xy  2yz) (5b)
xyz  l(2xz  2yz) (5c)
From Equations (5a) and (5b), we obtain
l(xy  2xz)  l(xy  2yz) (6)
Observe that l  0; otherwise, Equations (4a), (4b), and (4c) would imply that
yz  xz  xy  0, thus contradicting Equation (4d). Dividing both sides of (6) by l
and simplifying give
2xz  2yz or 2z(x  y)  0
Now z  0; otherwise, Equation (4a) would imply that l  0, which, as was observed
earlier, is impossible. Therefore, x  y.
Next, from Equations (5b) and (5c), we have
l(xy  2yz)  l(2xz  2yz) (7)
Dividing both sides of Equation (7) by l and simplifying, we get
xy  2xz or x(y  2z)  0
Since x  0, we have y  2z. Finally, substituting x  y  2z into Equation (4d) gives
4z 2  4z 2  4z 2  48
or z  2 (we reject the negative root, since z must be positive). Therefore, x  y  4,
and the dimensions of the box are 4 ft 4 ft 2 ft. Its volume is 32 ft3.

Interpreting Our Results

Geometrically, this problem is one of finding the point on the surface xy  2xz 
2yz  48 at which f(x, y, z)  xyz has the largest value. The point (4, 4, 2) is precisely
the point at which the level surface xyz  f(4, 4, 2)  32 is tangent to the surface
xy  2xz  2yz  48.
The next example shows how the method of Lagrange multipliers can be used to
help find the absolute extreme values of a function on a closed, bounded set.
1136 Chapter 13 Functions of Several Variables

EXAMPLE 5 Find the absolute extreme values of f(x, y)  2x 2  y 2  2y  1 sub-

ject to the constraint x 2  y 2 4.

Solution The inequality x 2  y 2 4 defines the disk D, which is a closed, bounded

set with boundary given by the circle x 2  y 2  4. So, following the procedure given
in Section 13.8, we first find the critical number(s) of f inside D. Setting
fx (x, y)  4x  0
fy (x, y)  2y  2  2(y  1)  0
simultaneously gives (0, 1) as the only critical point of f in D.
Next, we find the critical numbers of f on the boundary of D using the method of
Lagrange multipliers. Writing t(x, y)  x 2  y 2  4, we have
§f(x, y)  4xi  2(y  1)j and §t(x, y)  2xi  2yj
The equation §f(x, y)  lt(x, y) and the constraint equation give the system
4x  2lx (8a)
2(y  1)  2ly (8b)
x y 4
2 2
(8c)
Equation (8a) gives
2x(l  2)  0
that is, x  0 or l  2. If x  0, then Equation (8c) gives y  2. Next, if l  2,
then Equation (8b) gives
2(y  1)  4y or y  1
in which case x  13. So f has the critical points (0, 2), (0, 2), ( 13, 1) and
( 13, 1) on the boundary of D.
Finally, we construct the following table.

(x, y) f(x, y) ⴝ 2x 2 ⴙ y 2 ⴚ 2y ⴙ 1
(0, 1) 0
( 13, 1) 10
( 13, 1) 10
(0, 2) 9
(0, 2) 1

From the table we see that f has an absolute minimum value of 0 attained at (0, 1) and
an absolute maximum value of 10 attained at (13, 1) and ( 13, 1).

Optimizing a Function Subject to Two Constraints

Some applications involve maximizing or minimizing an objective function f subject
to two or more constraints. Consider, for example, the problem of finding the extreme
values of f(x, y, z) subject to the two constraints
t(x, y, z)  k and h(x, y, z)  l
It can be shown that if f has an extremum at (a, b, c) subject to these constraints, then
there are real numbers (Lagrange multipliers) l and m such that
§f(a, b, c)  l§t(a, b, c)  m §h(a, b, c) (9)
 13.9 Lagrange Multipliers 1137

z h(x, y, z)  l Geometrically, we are looking for the extreme values of f(x, y, z) on the curve of
intersection of the level surfaces t(x, y, z)  k and h(x, y, z)  l. Condition (9) is a
¬ g
f statement that at an extremum point (a, b, c), the gradient of f must lie in the plane
determined by the gradient of t and the gradient of h. (See Figure 9.) The vector equa-
tion (9) is equivalent to three scalar equations. When combined with the two constraint
P
equations, this leads to a system of five equations that can be solved for the five
μ h unknowns x, y, z, l, and m.
C

0 g(x, y, z)  k
EXAMPLE 6 Find the maximum and minimum values of the function f(x, y, z) 
3x  2y  4z subject to the constraints x  y  2z  1 and x 2  y 2  4.
x y
FIGURE 9
Solution Write the constraint equations in the form
If f has an extreme value at P(a, b, c), t(x, y, z)  x  y  2z  1 and h(x, y, z)  x 2  y 2  4
then §f(a, b, c)  l §t(a, b, c) 
m §h(a, b, c). Then the equation §f(x, y, z)  l §t(x, y, z)  m §h(x, y, z) becomes
3i  2j  4k  l(i  j  2k)  m(2x i  2yj)
 (l  2mx)i  (l  2my)j  2lk
Equating like components and rewriting the constraint equations lead to the following
system of five equations in the five variables, x, y, z, l, and m:
3 l  2mx (10a)
2  l  2my (10b)
4  2l (10c)
x  y  2z  1 (10d)
x 2  y2  4 (10e)
From Equation (10c) we have l  2. Next, substituting this value of l into Equations
(10a) and (10b) gives
3  2  2mx or 1  2mx (11a)
and
2  2  2my or 4  2my (11b)
Solving Equations (11a) and (11b) for x and y gives x  1>(2m) and y  2>m. Substi-
tuting these values of x and y into Equation (10e) yields
1 2 2 2
a b a b 4
2m m
17
1  16  16m2 or m2 
16
Therefore, m  117>4, so x  2> 117 and y  8> 117. Using Equation (10d),
we have

(1  x  y)  a1  b
1 1 2 8
z
2 2 117 117

a1 b
1 6

2 117
1138 Chapter 13 Functions of Several Variables

The value of f at the point 1 117

2 8
, 117 , 12  117
3
2 is
b  2a b  4a  b2
2 8 1 3 34
3a  2(1  117)
117 117 2 117 117
and the value of f at the point 1 117
2
, 117
8
, 12  3
117 2 is
b  2a b  4a  b2
2 8 1 3 34
3a  2(1  117)
117 117 2 117 117
Therefore, the maximum value of f is 2(1  117) , and the minimum value of f is
2(1  117) .

13.9 CONCEPT QUESTIONS

1. What is a constrained maximum (minimum) value problem? y
g(x, y)  4 k  6 k  4
Illustrate with examples. k  2
2. Describe the method of Lagrange multipliers for finding the 4
extrema of f(x, y) subject to the constraint t(x, y)  c. State
the method for the case in which f and t are functions of 2
three variables.
3. The figure at the right shows the contour map of a function f 0 x
and the curve of the equation t(x, y)  4. Use the figure to 4 2 2 4
obtain estimates of the maximum and minimum values of f 2 k6
subject to the constraint t(x, y)  4. k4
4 k2

13.9 EXERCISES
In Exercises 1–4, use the method of Lagrange multipliers to find 12. f(x, y, z)  x  y  z; x 2  y2  z2  1
the extrema of the function f subject to the given constraint. 13. f(x, y, z)  x  2y  2z; x 2  2y 2  4z 2  1
Sketch the graph of the constraint equation and several level
curves of f. Include the level curves that touch the graph of the 14. f(x, y, z)  x 2  y 2  z 2; yx1
constraint equation at the points where the extrema occur. 1 2
15. f(x, y, z)  xyz; x 2  2y 2  z 6
1. f(x, y)  3x  4y; x 2  y2  1 2
2. f(x, y)  x 2  y 2; 2x  4y  5 16. f(x, y, z)  xy  xz; x 2  y2  z2  8
3. f(x, y)  x 2  y 2; xy  1
In Exercises 17–20, use the method of Lagrange multipliers to
4. f(x, y)  xy; x  y2  4
2
find the extrema of the function subject to the given constraints.

In Exercises 5–16, use the method of Lagrange multipliers to 17. f(x, y, z)  2x  y; x  y  z  1, y2  z2  9

find the extrema of the function f subject to the given constraint. 18. f(x, y, z)  x  y  z; x 2  y 2  1, xz2
5. f(x, y)  xy; 2x  3y  6 19. f(x, y, z)  yz  xz; xz  1, y z 1
2 2

6. f(x, y)  x 2  y 2; x 2  y2  1 20. f(x, y, z)  x 2  y 2  z 2; 2x  y  z  2,

7. f(x, y)  xy; x 2  4y 2  1 x  2y  3z  4

8. f(x, y)  8x  9y; 4x 2  9y 2  36 In Exercises 21–22, use the method of Lagrange multipliers to find
9. f(x, y)  x  xy  y 2; x 2  y 2  8
2
the extrema of the function subject to the inequality constraint.
10. f(x, y)  x 2  y 2; x 4  y4  1 21. f(x, y)  3x 2  2y 2  2x  1; x 2  y 2 9
11. f(x, y, z)  x  2y  z; x  4y  z  0
2 2
22. f(x, y)  x y; 4x  y
2 2 2
4

V Videos for selected exercises are available online at www.academic.cengage.com/login.

 13.9 Lagrange Multipliers 1139

23. Find the point on the plane x  2y  z  4 that is closest to 38. Maximizing Profit The total daily profit (in dollars) realized by
the origin. Weston Publishing in publishing and selling its dictionaries
24. Find the maximum and minimum distances from the origin is given by the profit function
to the curve 5x 2  6xy  5y 2  10  0. P(x, y)  0.005x 2  0.003y 2  0.002xy  14x  12y  200
25. Find the point on the plane x  2y  z  5 that is closest to where x stands for the number of deluxe editions and y
the point (2, 3, 1). denotes the number of standard editions sold daily. Weston’s
26. Find the points on the surface z 2  xy  x  4y  21 that management decides that publication of these dictionaries
are closest to the origin. What is the shortest distance from should be restricted to a total of exactly 400 copies per day.
the origin to the surface? How many deluxe copies and how many standard copies
27. Find the points on the surface xy 2z  4 that are closest to should be published each day to maximize Weston’s daily
the origin. What is the shortest distance from the origin to profit?
the surface? 39. Cobb-Douglas Production Function Suppose x units of labor and
28. Find three positive real numbers whose sum is 500 and y units of capital are required to produce
whose product is as large as possible. f(x, y)  100x 3>4y 1>4
29. Find the dimensions of a closed rectangular box of maximum units of a certain product. If each unit of labor costs $200
volume that can be constructed from 48 ft2 of cardboard. and each unit of capital costs $300 and a total of $60,000 is
30. Find the dimensions of an open rectangular box of maximum available for production, determine how many units of labor
volume that can be constructed from 12 ft2 of cardboard. and how many units of capital should be used to maximize
production.
31. An open rectangular box having a volume of 108 in3. is to
be constructed from cardboard. Find the dimensions of such 40. a. Find the distance between the point P(x 1, y1) and the
a box if the amount of cardboard used in its construction is line ax  by  c  0 using the method of Lagrange
to be minimized. multipliers.
b. Use the result of part (a) to find the distance between the
32. Find the dimensions of the rectangular box of maximum point (2, 1) and the line 2x  3y  6  0.
volume with faces parallel to the coordinate planes that can
be inscribed in the ellipsoid 41. Let f(x, y)  x  y and t(x, y)  x  x 5  y.
a. Use the method of Lagrange multipliers to find the
x2 y2 z2
  1 point(s) where f may have a relative maximum or relative
4 9 16 minimum subject to the constraint t(x, y)  1.
33. Solve the problem posed in Exercise 32 for the general case b. Plot the graph of t and the level curves of f(x, y)  k
of an ellipsoid with equation for k  2, 1, 0, 1, 2, using the viewing window
[4, 4] [4, 4]. Then use this to explain why the
x2 y2 z2
  1 point(s) found in part (a) does not give rise to a relative
a2 b2 c2 maximum or a relative minimum of f
where a, b, and c are positive real numbers. c. Verify the observation made in part (b) analytically.
34. Find the dimensions of the rectangular box of maximum 42. Let f(x, y)  x 2  y 2, and let t(x, y)  x  y.
volume lying in the first octant with three of its faces lying a. Show that f has no maximum or minimum values when
in the coordinate planes and one vertex lying in the plane subjected to the constraint t(x, y)  1.
2x  3y  z  6. What is the volume of the box? b. What happens when you try to use the method of
Lagrange multipliers to find the extrema of f subject to
35. Solve the problem posed in Exercise 34 for the general case t(x, y)  1? Does this contradict Theorem 1?
of a plane with equation
43. Find the point on the line of intersection of the planes
x y z
  1 x  2y  3z  9 and 2x  3y  z  4 that is closest to the
a b c origin.
where a, b, and c are positive real numbers. 44. Find the shortest distance from the origin to the curve with
3
36. An open rectangular box is to have a volume of 12 ft . If the equation y  (x  1)3>2. Explain why the method of
material for its base costs three times as much (per square Lagrange multipliers fails to give the solution.
foot) as the material for its sides, what are the dimensions of 45. a. Find the maximum distance from the origin to the
the box that can be constructed at the minimum cost? Folium of Descartes, x 3  y 3  3axy  0, where a  0,
37. A rectangular box is to have a volume of 16 ft3. If the mate- x  0 and y  0, using symmetry.
rial for its base costs twice as much (per square foot) as the b. Verify the result of part (a), using the method of
material for its top and sides, find the dimensions of the box Lagrange multipliers.
that can be constructed at the minimum cost.