4100 AWL/Thomas_ch04p244-324 8/20/04 9:02 AM Page 278

278 Chapter 4: Applications of Derivatives

4.5 Applied Optimization Problems

To optimize something means to maximize or minimize some aspect of it. What are the di-
mensions of a rectangle with fixed perimeter having maximum area? What is the least ex-
pensive shape for a cylindrical can? What is the size of the most profitable production
run? The differential calculus is a powerful tool for solving problems that call for maxi-
x mizing or minimizing a function. In this section we solve a variety of optimization prob-
lems from business, mathematics, physics, and economics.

12
Examples from Business and Industry
EXAMPLE 1 Fabricating a Box
x
An open-top box is to be made by cutting small congruent squares from the corners of a
x x
12 12-in.-by-12-in. sheet of tin and bending up the sides. How large should the squares cut
from the corners be to make the box hold as much as possible?
(a)

Solution We start with a picture (Figure 4.32). In the figure, the corner squares are x in.
on a side. The volume of the box is a function of this variable:
Vsxd = xs12 - 2xd2 = 144x - 48x 2 + 4x 3 . V = hlw
x
12  2x Since the sides of the sheet of tin are only 12 in. long, x … 6 and the domain of V is the in-
12 terval 0 … x … 6.
12  2x A graph of V (Figure 4.33) suggests a minimum value of 0 at x = 0 and x = 6 and a
x
x maximum near x = 2. To learn more, we examine the first derivative of V with respect to x:
(b) dV
= 144 - 96x + 12x 2 = 12s12 - 8x + x 2 d = 12s2 - xds6 - xd.
dx
FIGURE 4.32 An open box made by
cutting the corners from a square sheet of Of the two zeros, x = 2 and x = 6, only x = 2 lies in the interior of the function’s domain
tin. What size corners maximize the box’s and makes the critical-point list. The values of V at this one critical point and two end-
volume (Example 1)? points are
Critical-point value: Vs2d = 128
Maximum
Endpoint values: Vs0d = 0, Vs6d = 0.
y 3
The maximum volume is 128 in. . The cutout squares should be 2 in. on a side.
y
x(12 – 2x)2,
0
x
6
Volume

EXAMPLE 2 Designing an Efficient Cylindrical Can

You have been asked to design a 1-liter can shaped like a right circular cylinder (Figure
min min 4.34). What dimensions will use the least material?
x
0 2 6 Solution Volume of can: If r and h are measured in centimeters, then the volume of the
NOT TO SCALE
can in cubic centimeters is
FIGURE 4.33 The volume of the box in pr 2h = 1000. 1 liter = 1000 cm3
Figure 4.32 graphed as a function of x.
2
Surface area of can: A = 2pr
()*
+ 2prh
()*
circular circular
ends wall

Copyright © 2005 Pearson Education, Inc., publishing as Pearson Addison-Wesley

4100 AWL/Thomas_ch04p244-324 8/20/04 9:02 AM Page 279

4.5 Applied Optimization Problems 279

2r How can we interpret the phrase “least material”? First, it is customary to ignore the thickness
of the material and the waste in manufacturing. Then we ask for dimensions r and h that make
the total surface area as small as possible while satisfying the constraint pr 2h = 1000.
To express the surface area as a function of one variable, we solve for one of the vari-
ables in pr 2h = 1000 and substitute that expression into the surface area formula. Solving
h for h is easier:
1000
h = .
pr 2
Thus,
FIGURE 4.34 This 1-L can uses the least A = 2pr 2 + 2prh
material when h = 2r (Example 2).
= 2pr 2 + 2pr a b
1000
pr 2
2000
= 2pr 2 + r .
Our goal is to find a value of r 7 0 that minimizes the value of A. Figure 4.35 suggests
that such a value exists.

Tall and
thin can
Short and
wide can

A
2␲r 2  2000
—— , r  0
r
Tall and thin
min

r
0 3 500
␲
Short and wide

FIGURE 4.35 The graph of A = 2pr 2 + 2000>r is concave up.

Notice from the graph that for small r (a tall thin container, like a piece of pipe), the
term 2000>r dominates and A is large. For large r (a short wide container, like a pizza
pan), the term 2pr 2 dominates and A again is large.
Since A is differentiable on r 7 0, an interval with no endpoints, it can have a mini-
mum value only where its first derivative is zero.
dA 2000
= 4pr -
dr r2
2000
0 = 4pr - Set dA>dr = 0 .
r2
4pr 3 = 2000 Multiply by r 2 .

r = 3 500 L 5.42 Solve for r.

A p
What happens at r = 2
3
500>p?

Copyright © 2005 Pearson Education, Inc., publishing as Pearson Addison-Wesley

4100 AWL/Thomas_ch04p244-324 8/20/04 9:02 AM Page 280

280 Chapter 4: Applications of Derivatives

The second derivative

d 2A 4000
2
= 4p +
dr r3
is positive throughout the domain of A. The graph is therefore everywhere concave up and
the value of A at r = 2 3
500>p an absolute minimum.
The corresponding value of h (after a little algebra) is
1000 500
h = = 2 3 p = 2r.
pr 2 A
The 1-L can that uses the least material has height equal to the diameter, here with
r L 5.42 cm and h L 10.84 cm.

Solving Applied Optimization Problems

1. Read the problem. Read the problem until you understand it. What is given?
What is the unknown quantity to be optimized?
2. Draw a picture. Label any part that may be important to the problem.
3. Introduce variables. List every relation in the picture and in the problem as
an equation or algebraic expression, and identify the unknown variable.
4. Write an equation for the unknown quantity. If you can, express the un-
known as a function of a single variable or in two equations in two un-
knowns. This may require considerable manipulation.
5. Test the critical points and endpoints in the domain of the unknown. Use
what you know about the shape of the function’s graph. Use the first and sec-
ond derivatives to identify and classify the function’s critical points.

Examples from Mathematics and Physics

EXAMPLE 3 Inscribing Rectangles
A rectangle is to be inscribed in a semicircle of radius 2. What is the largest area the rec-
tangle can have, and what are its dimensions?
y
Solution Let sx, 24 - x 2 d be the coordinates of the corner of the rectangle obtained by
x 2  y2
4 placing the circle and rectangle in the coordinate plane (Figure 4.36). The length, height,
x, 兹4  x 2 and area of the rectangle can then be expressed in terms of the position x of the lower
 
right-hand corner:
Length: 2x, Height: 24 - x 2, Area: 2x # 24 - x 2 .
2
Notice that the values of x are to be found in the interval 0 … x … 2, where the selected
x
–2 –x 0 x 2 corner of the rectangle lies.
Our goal is to find the absolute maximum value of the function
Asxd = 2x24 - x 2
FIGURE 4.36 The rectangle inscribed in
the semicircle in Example 3. on the domain [0, 2].

Copyright © 2005 Pearson Education, Inc., publishing as Pearson Addison-Wesley

4100 AWL/Thomas_ch04p244-324 8/20/04 9:02 AM Page 281

4.5 Applied Optimization Problems 281

The derivative

dA -2x 2
= + 224 - x 2
dx 24 - x 2
is not defined when x = 2 and is equal to zero when

-2x 2
+ 224 - x 2 = 0
24 - x 2
-2x 2 + 2s4 - x 2 d = 0
8 - 4x 2 = 0
x2 = 2 or x = ; 22.
Of the two zeros, x = 22 and x = - 22, only x = 22 lies in the interior of A’s do-
main and makes the critical-point list. The values of A at the endpoints and at this one crit-
ical point are
Critical-point value: A A 22 B = 22224 - 2 = 4
Endpoint values: As0d = 0, As2d = 0.
The area has a maximum value of 4 when the rectangle is 24 - x 2 = 22 units high and
2x = 222 unit long.
HISTORICAL BIOGRAPHY
Willebrord Snell van Royen EXAMPLE 4 Fermat’s Principle and Snell’s Law
(1580–1626)
The speed of light depends on the medium through which it travels, and is generally slower
in denser media.
Fermat’s principle in optics states that light travels from one point to another along a
path for which the time of travel is a minimum. Find the path that a ray of light will follow
in going from a point A in a medium where the speed of light is c1 to a point B in a second
medium where its speed is c2 .

Solution Since light traveling from A to B follows the quickest route, we look for a path
that will minimize the travel time. We assume that A and B lie in the xy-plane and that the
line separating the two media is the x-axis (Figure 4.37).
y
In a uniform medium, where the speed of light remains constant, “shortest time”
means “shortest path,” and the ray of light will follow a straight line. Thus the path from A
Angle of
A to B will consist of a line segment from A to a boundary point P, followed by another line
incidence
␪1 Medium 1 segment from P to B. Distance equals rate times time, so
a ␪1
P distance
x Time = rate .
0 x d
Angle of
b ␪2 refraction The time required for light to travel from A to P is
Medium 2

dx
B AP 2a 2 + x 2
t1 = c1 = c1 .

FIGURE 4.37 A light ray refracted

From P to B, the time is
(deflected from its path) as it passes
from one medium to a denser medium PB 2b 2 + sd - xd2
(Example 4).
t2 = c2 = c2 .

Copyright © 2005 Pearson Education, Inc., publishing as Pearson Addison-Wesley

4100 AWL/Thomas_ch04p244-324 8/20/04 9:02 AM Page 282

282 Chapter 4: Applications of Derivatives

The time from A to B is the sum of these:

2a 2 + x 2 2b 2 + sd - xd2
t = t1 + t2 = c1 + c2 .

This equation expresses t as a differentiable function of x whose domain is [0, d]. We want
to find the absolute minimum value of t on this closed interval. We find the derivative
dt x d - x
= - .
dx c1 2a + x
2 2
c2 2b + sd - xd2
2

In terms of the angles u1 and u2 in Figure 4.37,

dt sin u1 sin u2
= c1 - c2 .
dx
If we restrict x to the interval 0 … x … d, then t has a negative derivative at x = 0 and a
positive derivative at x = d. By the Intermediate Value Theorem for Derivatives (Section
dt/dx dt/dx dt/dx 3.1), there is a point x0 H [0, d] where dt>dx = 0 (Figure 4.38). There is only one such
negative zero positive point because dt>dx is an increasing function of x (Exercise 54). At this point
      sin u1 sin u2
x
x0 c1 = c2 .
0 d
This equation is Snell’s Law or the Law of Refraction, and is an important principle in
FIGURE 4.38 The sign pattern of dt>dx
the theory of optics. It describes the path the ray of light follows.
in Example 4.

Examples from Economics

In these examples we point out two ways that calculus makes a contribution to economics.
The first has to do with maximizing profit. The second has to do with minimizing average
cost.
Suppose that
rsxd = the revenue from selling x items
csxd = the cost of producing the x items
psxd = rsxd - csxd = the profit from producing and selling x items.

The marginal revenue, marginal cost, and marginal profit when producing and selling x
items are
dr
= marginal revenue,
dx
dc
= marginal cost,
dx
dp
= marginal profit.
dx
The first observation is about the relationship of p to these derivatives.
If r(x) and c(x) are differentiable for all x 7 0, and if psxd = rsxd - csxd has a
maximum value, it occurs at a production level at which p¿sxd = 0. Since p¿sxd =
r¿sxd - c¿sxd, p¿sxd = 0 implies that
r¿sxd - c¿sxd = 0 or r¿sxd = c¿sxd.

Copyright © 2005 Pearson Education, Inc., publishing as Pearson Addison-Wesley

4100 AWL/Thomas_ch04p244-324 8/20/04 9:02 AM Page 283

4.5 Applied Optimization Problems 283

Therefore

At a production level yielding maximum profit, marginal revenue equals mar-

ginal cost (Figure 4.39).

Cost c(x)

Dollars
Revenue r(x)

Break-even point
Maximum profit, c'(x)
r'(x)
B

Local maximum for loss (minimum profit), c'(x)
r'(x)

x
0 Items produced

FIGURE 4.39 The graph of a typical cost function starts concave down and later turns concave up. It
crosses the revenue curve at the break-even point B. To the left of B, the company operates at a loss. To
the right, the company operates at a profit, with the maximum profit occurring where c¿sxd = r¿sxd .
Farther to the right, cost exceeds revenue (perhaps because of a combination of rising labor and
material costs and market saturation) and production levels become unprofitable again.

EXAMPLE 5 Maximizing Profit

Suppose that rsxd = 9x and csxd = x 3 - 6x 2 + 15x, where x represents thousands of
units. Is there a production level that maximizes profit? If so, what is it?
y
Solution Notice that r¿sxd = 9 and c¿sxd = 3x 2 - 12x + 15.
c(x)
x 3  6x2  15x 3x 2 - 12x + 15 = 9 Set c¿sxd = r¿sxd.
3x 2 - 12x + 6 = 0
The two solutions of the quadratic equation are
12 - 272
r(x)
9x x1 = = 2 - 22 L 0.586 and
6
Maximum
for profit 12 + 272
x2 = = 2 + 22 L 3.414.
6

Local maximum for loss The possible production levels for maximum profit are x L 0.586 thousand units or x L
x 3.414 thousand units. The second derivative of psxd = rsxd - csxd is p–sxd = -c–sxd
0 2  兹2 2 2  兹2
NOT TO SCALE
since r–sxd is everywhere zero. Thus, p–sxd = 6s2 - xd which is negative at x = 2 + 22
and positive at x = 2 - 22. By the Second Derivative Test, a maximum profit occurs at
FIGURE 4.40 The cost and revenue about x = 3.414 (where revenue exceeds costs) and maximum loss occurs at about
curves for Example 5. x = 0.586. The graph of r(x) is shown in Figure 4.40.

Copyright © 2005 Pearson Education, Inc., publishing as Pearson Addison-Wesley

4100 AWL/Thomas_ch04p244-324 8/20/04 9:02 AM Page 284

284 Chapter 4: Applications of Derivatives

y EXAMPLE 6 Minimizing Costs

c(x)
5000
x  25x
A cabinetmaker uses plantation-farmed mahogany to produce 5 furnishings each day.
y
25x Each delivery of one container of wood is $5000, whereas the storage of that material is
$10 per day per unit stored, where a unit is the amount of material needed by her to pro-
duce 1 furnishing. How much material should be ordered each time and how often should
Cost

the material be delivered to minimize her average daily cost in the production cycle be-
tween deliveries?
y
5000
x
Solution If she asks for a delivery every x days, then she must order 5x units to have
enough material for that delivery cycle. The average amount in storage is approximately
x
x min one-half of the delivery amount, or 5 x>2. Thus, the cost of delivery and storage for each
Cycle length cycle is approximately
FIGURE 4.41 The average daily cost c(x) Cost per cycle = delivery costs + storage costs
is the sum of a hyperbola and a linear
a b
5x # #
function (Example 6). Cost per cycle = 5000 + x 10
()* 2 ()* ()*
delivery ()* number of storage cost
cost average days stored per day
amount stored
We compute the average daily cost c(x) by dividing the cost per cycle by the number of
days x in the cycle (see Figure 4.41).
5000
csxd = x + 25x, x 7 0.

As x : 0 and as x : q , the average daily cost becomes large. So we expect a minimum

to exist, but where? Our goal is to determine the number of days x between deliveries that
provides the absolute minimum cost.
We find the critical points by determining where the derivative is equal to zero:
5000
c¿sxd = - + 25 = 0
x2
x = ; 2200 L ;14.14.
Of the two critical points, only 2200 lies in the domain of c(x). The critical-point value of
the average daily cost is

c A 2200 B =
5000
+ 25 2200 = 50022 L $707.11.
2200
We note that c(x) is defined over the open interval s0, q d with c–sxd = 10000>x 3 7 0.
Thus, an absolute minimum exists at x = 2200 L 14.14 days.
The cabinetmaker should schedule a delivery of 5s14d = 70 units of the exotic wood
every 14 days.

In Examples 5 and 6 we allowed the number of items x to be any positive real number.
In reality it usually only makes sense for x to be a positive integer (or zero). If we must
round our answers, should we round up or down?

EXAMPLE 7 Sensitivity of the Minimum Cost

Should we round the number of days between deliveries up or down for the best solution in
Example 6?

Copyright © 2005 Pearson Education, Inc., publishing as Pearson Addison-Wesley

4100 AWL/Thomas_ch04p244-324 8/20/04 9:02 AM Page 285

4.5 Applied Optimization Problems 285

Solution The average daily cost will increase by about $0.03 if we round down from
14.14 to 14 days:
5000
cs14d = + 25s14d = $707.14
14
and
cs14d - cs14.14d = $707.14 - $707.11 = $0.03.
On the other hand, cs15d = $708.33, and our cost would increase by $708.33 -
$707.11 = $1.22 if we round up. Thus, it is better that we round x down to 14 days.

Copyright © 2005 Pearson Education, Inc., publishing as Pearson Addison-Wesley