Section 3.

7 Optimization
An optimization problem is essentially a word problem which asks you to
maximize or minimize a certain quantity. Nearly all optimization problems
can be solved by employing a fixed strategy. It goes something like this:

• Read and re-read the problem until you understand it. In particular,
make sure you know the quantity you are being asked to maximize or
minimize.

• Draw and label a picture which gives the relevant information.

• Write equations that describe

(a) the quantity you are attempting to maximize/minimize in terms
of other variables which appear in your drawing, and
(b) the constraints that your variables must obey

• Solve Equation in part (b) above for one of the variables, and plug
this result back into Equation in part (a).

• Determine the ”domain” of your function. This is very important!

• Find the extrema

either using the Extreme Value Theorem(EVT) (if your domain is a
closed interval)
or use One-critical-value Extrema Theorem (to be introduced in this
section later)

1
 Optimization on closed intervals

Example A farmer has a 2,400 ft of fencing and wants to fence off a rect-
angular field that borders a straight river. (No fence needed along the river
side). What are the dimensions of the field that has the largest area?

Here is my solutions process: After I have read the problems a few times to
understand what it’s asking I know that I’m asked to maximize the area(A)
of a rectangular region. I also know that the total fence length is 2,400 ft. I
do not know the length and the width of the rectangular field.

Then I like to draw a picture of whats going on. In the picture Ill label cer-
tain quantities so I can keep track of whats happening. Below W represents
the width and L represents the length of the field

I will maximize the area and the area of a rectangle is A = L · W . Perhaps

a less obvious thing to write down is what I call a constraining equation;
its an equation that remembers how the variables W and L relate to each
other. In this case, the relationship is: 2W + L = 2400
Because the total fencing available will be used to cover the three sides of
this farmland. The total length of these sides is W+W+L.
What I’d like to do next is to take my expression for area and optimize it.
The problem is that; currently it sits as a function of two variables instead of
one. We don’t have the technology yet to optimize a function of 2 variables.
So what do we do? We will take our constraining equation and use it to
solve for L in terms of W: 2W + L = 2, 400 ⇒ L = 2, 400 − 2W .

2
Now I can substitute this value for L back into my expression for area to
come up with an expression that makes area a function of one variable

A(W ) = (2, 400 − 2W ) · W = 2, 400W − 2W 2

Next the important thing is to find the ”domain” of this function. Naturally
since this is a polynomial I believe most of you will say it is all real numbers.
But in word problems such as this one you not only have to take the math
into consideration but also the nature of the problem. That is why the word
”domain” is in quotes in the recipe and here. What do I mean by the nature
of the problem? Here since my variable W represents a length it does not
make sense to consider the case when W < 0. Similarly, if W > 1, 200 then
the other dimension of the rectangle, L, will be negative. So the ”domain”
I’m trying to maximize the area on is [0, 1200].

Since A(W) is a polynomial it is continuous anywhere hence on the in-

terval [0, 1200] so I may use the EVT we have learnt in Section 3.3 for
A(W ) = 2400W − 2W 2 on the interval [0, 1200]. First find the critical val-
ues then :

A0 (W ) = 2400 − 4W = 0 ⇒ W = 600
Then I need the comparison chart below to find the Absolute Maximum

W A(W)
0 A(0)=0
600 A(600)=720,000
1200 A(1200)=0
So the maximum area is then 720, 000f t2 and the dimensions of this field
are W = 600ft, L = 2400 − 2(600) = 1200ft.

Example A piece of wire 12m long is cut into two pieces. One piece is bent
into a square and the other is bent into an equilateral triangle. How should
the wire be cut so that the total area enclosed is (a) a minimum and (b) a
maximum? (Note: it is allowed to not cut the wire at all, but instead to use
all of it to construct either a square or a triangle.)

Again after I’ve read the problems a few times to understand what it’s ask-
ing. I know I have to maximize (minimize) the area again. To find the
equations I like to draw a picture of what’s going on. In the picture I’ll
label certain quantities so I can keep track of what’s happening. For this

3
problem, I’ll draw my piece of wire cut into two pieces (leaving two pieces
of length u and y) and what these pieces of wire look like after they have
been fashioned into a square and a triangle.

We can see from the picture that the total area enclosed by both figures is
√ √
u 2 3 y 2 u2 3 2
A = A1 + A2 = ( ) + ( ) = + y
4 4 3 16 36
(the second term I know by using the pythagorean theorem to solve for the
height of the triangle). In this problem, the constraining equation is given
by u + y = 12

What I’d like to do is take my expression for area and optimize it. The
problem is again it sits as a function of two variables instead of one. Once
again take our constraining equation and use it to solve for y in terms of u:

u + y = 12 ⇒ y = 12 − u
Now substitute this value for y back into my expression for area to come up
with an expression that makes area a function of one variable :
√
u2 3
A(u) = + (12 − u)2
16 36
I have area as a function of the variable u. Notice that u can vary between
0 and 12, so were now asking the following problem: find the absolute
minimum (part (a)) and absolute maximum (part (b)) of the function A(u)
on the interval [0, 12]. How do we solve this problem? First, we verify that
A(u) is a continuous function on the closed interval [0, 12]. Since it is a

4
polynomial, it certainly is continuous. so we may use the EVT. So let’s find
critical points for A(u). Since
√
0 1 3
A (u) = u − (12 − u)
8 18
we see first that the derivative is defined everywhere so that critical points
are simply
√
zeroes
√
of the first derivative. When is A0 (u) = 0?
1 3 12 3
8 u + 18 u = 18 ⇒ u ≈ 5.21

Finally, we evaluate the function at these critical points to find the maximum
and minimum
u A(u)
0 A(0) ≈ 6.93
5.21 A(5.21) ≈ 3.92
12 A(12) ≈ 9
From the table we see that area is maximized when u = 12 (that is, when
all of the wire is used to make the square) and minimized when 5.21 meters
is used to make the square (and so 6.79 meters is used to make a triangle).

Optimization on open intervals

It often happens that you are asked to optimize a function on an open

interval (or the whole real line) instead of a closed interval? How does one go
about solving such a problem? Happily, much of the technique in optimizing
functions on closed intervals will carry over to optimizing functions on open
intervals. The main difference is that in finding absolute maxima or minima
we will use a variation of the 1st derivative test instead of our techniques
for finding absolute maxima and minima for continuous functions on closed
intervals. This variation of the 1st derivative test is:

5
One-Critical-Value Extrema Theorem Suppose f(x) is a continuous
function defined on (a, b) (where here a = −∞ and b = ∞ are possible). If
f 0 (c) = 0 or undefined and f 0 (x) < 0 for x < c and f 0 (x) > 0 for x > c then
f has an absolute minimum (in its domain) at c. If f 0 (c) = 0 or undefined
and f 0 (x) > 0 for x < c and f 0 (x) < 0 for x > c then f has an absolute
maximum (in its domain) at c.

Remarks 1) Notice that this is very similar to the 1st derivative test
for local extrema. The difference is that in this test we require information
about the sign of the derivative for all x < c and x > c; in the local test, we
needed to only know this information for values of x near c.

2) To use this theorem for absolute extrema directly, you should be in the
situation where you have exactly one critical point in your domain.
If you have more than one critical point in your domain, you probably have
to do more work.

Example Find two numbers whose sum is 23 and whose product is maxi-
mum.

Normally we would start a problem like this drawing a picture. For this
problem, though, there is no real need. We see were looking for numbers
x and y such that x + y = 23 and so that their product P is maximized:
P = x · y. Notice in particular that x and y have no further restriction.
In particular, x could be any real number. Now since x + y = 23, we have
y = 23 − x. Hence were trying to maximize

P (x) = x(23 − x) = 23x − x2

on the domain (−∞, ∞). So we will use One-critical-value-extrema theorem.

P 0 (x) = 23 − 2x = 0 ⇒ x = 23 2
To find the sign of the derivative for values x < 23/2 it is enough to compute
the sign of the derivative for one value of x < 23/2. Similarly, to test the
sign of the derivative for values x > 23/2 it is enough to compute the sign of
the derivative for one value x > 23/2. So we may use our sign chart below

6
So we got x = 23/2 is a local maximum of P (x), as desired which gives us
an absolute maximum for P at x = 23/2 by the theorem. Hence, the two
numbers we are looking for are x = 23 23
2 and y = 23 − 2 = 2 .
23

Handy rule of thumb More often than not, if you are asked to optimize
area (with fixed perimeter) or volume (with fixed surface area), youll be
working on a closed interval. If you are asked to optimize perimeter (with
fixed area) or surface area (with fixed volume), you are probably working
on an open interval.

Example A cylindrical can is to be made to hold a 1 liter of oil. Find the

dimensions that will minimize the cost of the metal to manufacture the can.

We are asked to minimize the dimensions, this means in a way you are asked
to minimize the surface area of the can.And here is a picture of this can with
radius r and height h

7
Let S denote the surface area of this cylindrical can then
S = 2π · r2 + 2π · r · h

because the top and bottom of can have surface area of π · r2 each because
they are circles and the second term is the surface area of the side of the
can which is a rectangle with width = h and length = 2π · r. The constraint
equation for this problem is the volume info we are given, volume of the can
–which we will denote by V– is 1000cm3 (=1 liter). Also I know by the front
of the book the volume of a cylinder is V = h · π · r2 implies the constraint
equation h · π · r2 = 1000. Now solve this equation for h: h = 1000
πr2
and plug
it into the surface area equation to turn S into a function of one variable:
1000 2000
S(r) = 2π · r2 + 2π · r · ( 2
) = 2πr2 +
πr r
The only sure restriction for the variable radius(r) I can come up with is;
r > 0. So we are considering to maximize the surface area over the domain
(0, ∞). The handy rule above also suggested that with these kind of prob-
lems (fixed volume/optimize surface area) this kind of domain is no surprise.
So I’ll apply the One-critical value theorem again since my S is continuous
on this interval.So let’s find the critical value(s): q
2000 2000 500 500
S 0 (x) = 4πr = r2
= 0 ⇒ 4πr = r2
⇒ r3 = π ⇒r= 3
3

Finding only one critical value is promising. But I still have to show the
sign change at this r to conclude that it gives me the absolute minimum
for the surface area. I’ll leave
q this to you as an exercise to make the sign
chart and show that at r = 3 500
3 S has a minimum. Hence when the radius
q
is r = 3 500
3 ≈ 5.42cm S has an absolute minimum. Then the height is
1000
h= q ≈ 58.74cm
π( 3 500
3
)2