Minimizing the Calculus in

Optimization Problems
Teylor Greff

Mathematics Department
Whitman College
May 2016
Abstract: Do we actually need calculus to solve maximum/minimum
problems? Optimization problems are explored and solved using the AM/GM
inequality and Cauchy Schwarz inequality, while simultaneously finding trends
and evolutions in these optimization problems as we look at a textbooks
ranging from 1902 − 2015.

Contents
1 Introduction 3

2 The AM/GM Inequality 3

2.1 Diagram Proof . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.2 Algebra Proof . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.3 Formal Proof . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

3 Optimization Problems and the AM/GM Inequality 7

3.1 Fence Problems . . . . . . . . . . . . . . . . . . . . . . . . . . 7
3.1.1 Derivative Approach . . . . . . . . . . . . . . . . . . . 8
3.1.2 AM/GM Approach . . . . . . . . . . . . . . . . . . . . 8
3.1.3 Similar Problem Examples . . . . . . . . . . . . . . . . 9
3.2 Number Problems . . . . . . . . . . . . . . . . . . . . . . . . . 9
3.2.1 Incorrect Application . . . . . . . . . . . . . . . . . . . 10
3.2.2 Correct Application . . . . . . . . . . . . . . . . . . . . 10
3.2.3 Similar Problem Examples . . . . . . . . . . . . . . . . 11
3.3 Geometry Problems . . . . . . . . . . . . . . . . . . . . . . . . 11
3.3.1 Application . . . . . . . . . . . . . . . . . . . . . . . . 11
3.3.2 Similar Problem Examples . . . . . . . . . . . . . . . . 13
3.4 Three Dimensional Problems . . . . . . . . . . . . . . . . . . . 13
3.4.1 Application . . . . . . . . . . . . . . . . . . . . . . . . 13
3.4.2 Similar Problem Examples . . . . . . . . . . . . . . . . 14

4 Non-AM/GM Generalized Problems 14

4.1 “Triathlon” Problems . . . . . . . . . . . . . . . . . . . . . . . 14
4.1.1 Specific Solution . . . . . . . . . . . . . . . . . . . . . 15
4.1.2 Generalized Solution . . . . . . . . . . . . . . . . . . . 16
4.1.3 Similar Problem Examples . . . . . . . . . . . . . . . . 17
4.2 Pipe Corridor Problem . . . . . . . . . . . . . . . . . . . . . . 18
4.2.1 Generalized Solution . . . . . . . . . . . . . . . . . . . 18

1
4.2.2 Specific Solution . . . . . . . . . . . . . . . . . . . . . 19
4.2.3 Similar Problem Examples . . . . . . . . . . . . . . . . 19
4.3 Post Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
4.3.1 Generalized Solution . . . . . . . . . . . . . . . . . . . 20
4.3.2 Specific Solution . . . . . . . . . . . . . . . . . . . . . 21
4.3.3 Similar Problem Examples . . . . . . . . . . . . . . . . 21

5 Problems that Have Evolved 21

5.1 Number Problem → Box Problem . . . . . . . . . . . . . . . . 22
5.1.1 Similar Problem Examples . . . . . . . . . . . . . . . . 23
5.2 Graph Problem → Pipe Problem → Ladder Problem . . . . . 23
5.2.1 Similar Problem Examples . . . . . . . . . . . . . . . . 24

6 The CS Inequality 25
6.1 Formal Proof - CS . . . . . . . . . . . . . . . . . . . . . . . . 25

7 Optimization Problems and the CS Inequality 26

7.1 Typical Problem . . . . . . . . . . . . . . . . . . . . . . . . . 26
7.2 Optimization Problem . . . . . . . . . . . . . . . . . . . . . . 27

8 Historical Bit 29
8.1 Economic/Medical/Physical Problems . . . . . . . . . . . . . . 29

9 Conclusion 30

2
1 Introduction
The focus of this paper is optimization problems in single and multi-variable
calculus spanning from the years 1900 − 2016. The main goal was to see if
there was a way to solve most or all optimization problems without using any
calculus, and to see if there was a relationship between this discovery and
the published year of the optimization problems. The primary method used
to accomplish this was the Arithmetic Mean/Geometric Mean inequality,
but upon further exploration, the Cauchy Schwarz inequality and other
generalized solutions proved to be incredibly helpful as well. This paper
lays out how these problems are solved, gives examples of similar problems,
and also explores the historical evolution of these problems.

2 The AM/GM Inequality

The Arithmetic Mean/Geometric Mean inequality was the main method
when it came to solving optimization problems without differentiation. Before
looking at applications of the inequality, a firm grasp on the inequality
concept is key. The result states that the Arithmetic Mean is always greater
than or equal to the Geometric Mean. The AM/GM inequality is stated
below with three proofs. Two for intuition and one formal proof. We will
begin by looking at a diagram proof and an algebraic proof for two positive
numbers a and b before seeing the AM/GM applied to the nth case.

2.1 Diagram Proof

The following proof for the AM/GM inequality is useful for two numbers.
First, note that the arithmetic mean and geometric mean for two numbers
are as follows:
a+b √
AM = , GM = ab.
2
Using a semicircle where AB is the diameter representing a + b and P is the
point distinguishing a from b, we can illustrate the AM/GM.

Since the diameter of the circle is a + b, its radius is (a + b)/2. The red
line represents the arithmetic mean of a and b.

3
A C P B
a b

We now draw an altitude from point P to point D on the semicircle and

line segments from point D to points A and B. Thus, since ∆ADP is similar
to ∆DBP it follows that ∆ADB is a right triangle, since an angle inscribed
in a semicircle is a right angle, or:
a h √
= , h = ab.
h b

A C P B
a b

Therefore, the altitude is the geometric mean shown in blue, and is clearly
less than the hypotenuse.

4
D

AM GM

A C P B
a b

Therefore, no matter what the lengths of a and b, AM ≥ GM, and equality

holds whenever a = b.

2.2 Algebra Proof

Another way to easily reason the√ inequality
√ 2 with two positive numbers a and
b, is by first recognizing 0 ≤ ( a − b) , then it follows easily:
√ √
0 ≤ a − 2 a b + b,
√ a+b
ab ≤ .
2

2.3 Formal Proof

Now that we have proved the AM/GM inequality for two numbers, we prove
the inequality for n numbers. To prove this inequality, we use the following
lemma.
Lemma 1. Let n ≥ 2 be an integer. Suppose that b1 , b2 , b3 , ..., bn are positive
real numbers that are not all equal. If b1 b2 b3 · · · bn = 1, then b1 + b2 + b3 +
· · · + bn > n.
Proof. We will use the Principle of Mathematical Induction. For the case
n = 2 we know that b1 6= b2 and b1 b2 = 1. It follows that,
p p p
0 < ( b1 − b2 )2 = b1 − 2 b1 b2 + b2 = b1 − 2 + b2 and thus b1 + b2 > 2,

5
showing that the result is true when n = 2. Now suppose the result is
valid for some positive integer p ≥ 2. Let b1 , b2 , ..., bp , bp+1 be positive real
numbers that are not all equal and satisfy b1 b2 · · · bp bp+1 = 1. Without loss
of generality, we may assume that the numbers are in increasing order, that
is, b1 ≤ b2 ≤ · · · ≤ bp ≤ bp+1 . By the assumptions on these numbers, we
must have b1 < 1 < bp+1 . Since the conclusion of the lemma is assumed to
be true when n = p, we consider the product (b1 bp+1 )b2 · · · bp = 1, which is a
product of p numbers. If all of these numbers are equal (and thus all equal
1), then
b2 + b2 + · · · + bp = p − 1 and b1 + bp+1 > 2
(the inequality follows from the first part of the proof) and it follows that

b1 + b2 + · · · + bp + bp+1 > p + 1.

If the numbers b1 bp+1 , b2 , ..., bp are not all equal, then

b1 bp+1 + b2 + · · · + bp > p

by the induction hypothesis. Since the quantity (bp+1 − 1)(1 − b1 ) is positive,

we find that

b1 + b2 + · · · + bp+1 = (b1 bp+1 + b2 + · · · + bp ) + 1 + (bp+1 − 1)(1 − b1 )

> p + 1 + (bp+1 − 1)(1 − b1 )

> p + 1.
This shows that the result holds when n = p + 1. By the Principle of
Mathematical Induction, the conditional statement given in the lemma is
valid for all integers n ≥ 2. [7]
Theorem 2 (Arithmetic Mean/Geometric Mean Inequality). Let n be a
positive integer. If a1 , a2 , . . . , an are nonnegative real numbers, then
1 a1 + a2 + · · · + an
(a1 a2 · · · an ) n ≤ .
n
Equality occurs if and only if a1 = a2 = · · · = an .
Proof. The equality clearly holds when a1 = a2 = · · · = an . If n = 1, then
the result is trivial. Also, if any of the ak = 0, the result is trivial as well.

6
Now, suppose for n ≥ 2, all ak ’s are positive, and that not all of ak ’s are
1
equal to each other. If we let r = (a1 a2 ...an ) n , we get the following result,
a1 a2 an a1 a2 · · · an
··· = = 1.
r r r rn
By the lemma, it follows that,
a1 a2 an
+ + ··· + > n.
r r r
Multiplying through by r and dividing by n, we find that,
a1 + a2 + · · · + an 1
> r = (a1 a2 · · · an ) n . [7]
n
This completes the proof.

3 Optimization Problems and the AM/GM

Inequality
The following problems are typical problems seen in most calculus textbooks.
These can all be solved using the AM/GM inequality, and are categorized into
a few different types of problems that often appear in maximum/minimum
sections of calculus textbooks. Starting with a simple example, the derivative
approach is used, then a solution is shown using the AM/GM inequality. The
rest of the problems that follow, are solved using the AM/GM inequality.

3.1 Fence Problems

Fence problems are extremely popular in calculus textbooks, and perhaps
the most basic. There are many variations of this problem, but all can be
solved using the AM/GM inequality and provide an easy starting place to
see an application of the inequality.

Example 3. A landowner wishes to use 2 miles of fencing to enclose a

rectangular region of maximum area. However, one side runs along a stream,
so only three sides must be fenced in. Find the lengths of the sides of the
rectangular region having the largest area that can be enclosed. [5]

7
river

x x

3.1.1 Derivative Approach

We first solve the problem using calculus. Suppose that x is the width of
the rectangle, and y is its length. We want to maximize the area A, where
A = xy. The problem of maximizing the area of the region can be reduced
to the following:

maximize xy subject to 2x + y = 2.

Noticing that y = 2 − 2x and thus,

A(x) = x(2 − 2x) = 2x − 2x2 ,

we can take the derivative and set it equal to zero to find our maximal points:
1
0 = 2 − 4x and hence x = .
2
We have thus shown for the fencing to have maximum area, it will have a
width of 1/2 miles, and length of 1 mile.

3.1.2 AM/GM Approach

As above, we seek a solution to the following problem:

maximize xy subject to 2x + y = 2.

By the AM/GM inequality, we find that

p 1
2 = 2x + y ≥ 2 2xy and hence xy ≤ .
2
The maximum value for xy is thus (1/2) and this value is attained when

2x = 1 = y.

8
We have thus shown that the maximum area of the landowners fencing has
length of 1/2 miles, and width of 1 mile.

Fence problems, such as this one, give a good idea of how to apply the
AM/GM inequality to optimization problems. However, as we will see in the
next section, some care is needed when applying it, as not all problems work
out as cleanly as this one.

3.1.3 Similar Problem Examples

Example 4. A farmer wishes to set aside one acre of his land for corn and
wheat. To keep out the cows, the field is enclosed by a fence costing 50 per
running foot. In addition, a fence running down the middle of the field is
needed with a cost per foot of $1. Given that 1 acre = 43560 square feet,
what dimensions should the field have so as to minimize his total cost? The
field is rectangular. [1]
Example 5. A farmer wishes to divide 20 acres of land along a river into 6
smaller plots by using a one fence parallel to the river and 7 fences perpendicular
to the it. Show that the total amount of fencing is minimized if the sum of
the lengths of the 7 cross fences equals the length of the one fence parallel to
the river. [1]
Example 6. A dairy farmer plans to fence in a rectangular pasture adjacent
to a river. The pasture must contain 180, 000 square meters in order to
provide enough grass for the herd. What dimensions would require the least
amount of fencing if no fencing is needed along the river? [2]
Example 7. A rancher has 200 feet of fencing with which to enclose to two
adjacent rectangular corrals. What dimensions should be used so that the
enclosed area will be a maximum? [2]
Example 8. One side of an open field is bounded by a straight river. How
would you put a fence around the other three sides of a rectangular plot in
order to enclose as great an area as possible with a given length of fence? [4]

3.2 Number Problems

Number problems are extremely popular when it comes to early single variable
calculus textbooks. They involve finding two numbers that satisfy certain

9
conditions. The following example shows a problem that can occur when
using the AM/GM inequality. It often can appear unhelpful if used carelessly.
We will show what must occur in order for the AM/GM inequality to solve an
optimization problem correctly by first showing what goes wrong if applied
carelessly, and then showing how to carefully apply it.

Example 9. Divide the number 20 into two parts such that the product of
one part by the square of the other part shall be a maximum. [8]

3.2.1 Incorrect Application

Suppose that x is the part being squared, and y is the part being multiplied.
We want to maximize the product F , where F = x2 y. The problem of
maximizing the product of the numbers can now be reduced to the following:

maximize x2 y subject to x + y = 20.

By the AM/GM inequality, we find that

2
√ 20
20 = x + y ≥ 2 xy and hence xy ≤ = 100.
2

However, we notice that this states a maximum value for xy, when we are
looking for a maximum value of x2 y. This is where the inequality appears
unhelpful. Fortunately, there is a way to work around this as shown below.

3.2.2 Correct Application

Since we need two x values to appear, as that is what we are trying to
maximize, we write x + y as 21 x + 12 x + y. By the AM/GM inequality,
r 3
1 1 3 1 2 2 20
20 = x + x + y ≥ 3 x y and hence x y ≤ 4 .
2 2 4 3

We notice that this states the maximum value for x2 y is 32000/27 and this
value is attained when
1 20
x= = y.
2 3
Thus, we have shown that the maximum product of x2 y is when y = 20/3 ≈
13.33 and x = 40/3 ≈ 6.67.

10
This same idea will appear in many optimization problems when using the
AM/GM inequality. In order to solve these problems correctly, we must make
sure that the AM/GM is maximizing the quantity desired. The constraint
must multiply through to the get the value being maximized or vice versa.
In situations where both the constraint and the max/min function contain
addition, the AM/GM proves to be unhelpful. This will further be explored
in the Non-AM/GM Generalized Problems section.

3.2.3 Similar Problem Examples

Example 10. The product of two numbers is 16. Determine them so that
the square of one plus the cube of the other is as small as possible. [1]

Example 11. Divide the number 10 into two parts such that the square of
one part multiplied by the cube of the other shall be a maximum. [12]

3.3 Geometry Problems

In addition to number problems, geometry problems were a popular type of
optimization problem used in the early 1900’s. To this day almost all text
books include a problem much like the one below, and most can be solved
used the AM/GM as long as applied carefully as discussed above. Below is
an example and solution to one of these problems.

3.3.1 Application
Example 12. Determine the ratio of significant dimensions so as to attain
the maximum volume of a cone in a sphere. [6]

11
h

Suppose that R is the radius of the sphere. Let h and r be the height
and radius of the cone, respectively. We want to maximize the volume V
of the cone, where V = πr2 h/3. Referring to the figure, we note that h
will be larger than R when the volume of the cone is maximized. By the
Pythagorean Theorem,
(h − R)2 + r2 = R2 and thus h2 + r2 = 2Rh.
The problem of maximizing the volume of the cone can now be reduced to
the following:
2r2
maximize r2 h subject to 2h + = 4R.
h
By the AM/GM inequality, we find that
2r2 √
3
3
1 4R 32
2
4R = h + h + 2
≥ 3 2r h and hence r h ≤ = R3 .
h 2 3 27
The maximum value for r2 h is thus 32R3 /27 and this value is attained when
4R 2r2
h= = .
3 h
√
Since h2 = 2r2 , the ratio of height to radius for the optimal cone is 2. We
have thus shown that the maximum volume of a cone inscribed in a sphere
of radius R is
1 2 π 32 8 4 3
πr h = · R3 = · πR ,
3 3 27 27 3
that is, the maximal cone occupies 8/27 of the volume of the entire sphere.

12
3.3.2 Similar Problem Examples
Example 13. Find the maximum right cylinder that can be inscribed in a
sphere of radius a. [12]

3.4 Three Dimensional Problems

Not only can the AM/GM inequality be applied to single variable optimization
problems, but also multi-variable calculus problems. These problems are
typically solved using Lagrange Multipliers, but the AM/GM inequality gives
a clean and straightforward result. Below is an example of one of these
problems.

3.4.1 Application
Example 14. A rectangular building is being designed to minimize heat loss.
The east and the west walls lose heat at a rate of 10 units/m2 per day, the
north and south walls at a rate of 8 units/m2 per day, the floor at a rate of
1 unit/m2 , and the roof at a rate of 5 units/m2 per day. The volume must
be exactly 4000 m3 . Find the dimensions that minimize heat loss. [9]

u
5 m2

u
10 m2
h 8 u
m2 y

Suppose that x is the length of the box, y is the width of the box and h
is the height of the box. We want to minimize the heat loss function, f (x),
where f (x) = 20hy + 16xh + 6xy. The problem of minimizing the heat loss
can be reduced to the following:

minimize 20hy + 16xh + 6xy subject to xyh = 4000.

By the AM/GM inequality, we find that

p
20hy+16xh+6xy ≥ 3(1920x2 y 2 h2 )1/3 and hence 3 3 1920(4000)2 ≤ 20hy+16xh+6xy.

13
Since it follows that y = 8/3h from 20hy = 16xh = 6xy, the minimum value
is thus attained when
p
(20h)(8/3h) ≥ 3 1920(4000)2 .

Solving for h we find the optimal box to have h ≈ 7.6, y = 20.4 and x = 25.5.

Another interesting thing to note is the use of economic ideas in this calculus
problem. Looking through old books, economic and medical applications
did not begin appearing until around the 1970’s when calculus became a
requirement for other disciplines besides just mathematics. Further exploration
of this will be shown in the Historical Bit section of this paper.

3.4.2 Similar Problem Examples

Example 15. An open box is to be made from a rectangular piece of material
by cutting equal squares from each corner and turning up the sides. Find the
dimensions of the box of maximum volume if the material has dimensions 2
feet by 3 feet. [2]

Example 16. An open rectangular box is to be made from a piece of cardboard

8 in. wide and 15 in. long by cutting a square from each corner and bending
up the sides. Find the dimensions of the box of maximum volume. [4]

4 Non-AM/GM Generalized Problems

Many optimization problems cannot be solved using the AM/GM inequality.
I was able to find that if the equation being maximized/minimized and
the constraint both involved addition the AM/GM inequality proved to be
unhelpful. In these instances, calculus proved to be unavoidable, but a
convenient solution could be found to help with further similar problems.
Below are a few examples of these types of problems.

4.1 “Triathlon” Problems

“Triathlon” Problems take the form of a person or thing that needs to travel
in one direction at a certain speed, and in another direction at a different
speed. While the activity or subject may change, these problems can be

14
generalized to a simple solution. If a student was able to recognize a problem
as a “Triathalon” Problem, they could plug into the generalized solution
without using any calculus. One of these problems is shown below, first
using calculus, then using the generalized solution on a “different” problem.

Example 17. A man who can row at a speed of 4 miles per hour and run at
a speed of 6 miles per hour wishes to reach the point P from a boat at point
B as shown in the figure below in the least amount of time as possible. Find
the distance AP that the man must run on the beach. [8]

B
√ x
2

b
+
b2

C x P
A a−x

4.1.1 Specific Solution

Using the derivative, suppose that a is the distance from C to P , b is the
distance from B to C, and x is the distance from C to A. Recognizing,
distance=(rate)(time), then the problem of minimizing the time, t(x) can be
reduced to the following:
√
102 + x2 10 − x
t(x) = + where 0 ≤ x ≤ 10.
4 6
We can take the derivative and set it equal to zero to find our minimal points:
x 1
t0 (x) = √ − = 0,
2
4 10 + x 2 6

15
√
9x2 = 2 102 + x2 ,
5x2 = 400,
√
x = 80.
We have shown√to minimize the time to get from B to P is by traveling to
the point
√ A at 80 miles from point C. The distance from C to P is thus
10 − 80 ≈ 1.056 miles.

While this result is correct, we can make it much simpler for later similar
problems by generalizing the solution.

4.1.2 Generalized Solution

If we let the rate from point B to A be equal to r, and the rate from A to P
be equal to s, the minimum time t(x) is:
√
t(x) = r x2 + b2 + s(a − x).

Taking the derivative and setting equal to 0 gives,

x s
√ = ,
x 2 + b2 r

r 2 x2 = s 2 x2 + s 2 b 2 ,
s 2 b2
x2 = ,
r 2 − s2
sb
x= √ .
r 2 − s2
What is perhaps the most interesting result about this is that a does not
appear in our general equation, suggesting that after a certain point, no
matter how long a is, the person should still always arrive at the same point.
Also note r > s. Further, any problem set up in a similar manner can simply
applied with the above equation. An example of this is shown below.
Example 18. It is known that homing pigeons fly faster over land than over
water. Assume that they fly 10 meters per second over land, but only 8 meters
per second over water. If a pigeon is located at the edge of a straight river
500 meters wide and must fly to its nest, located 1300 meters away on the
opposite side of the river. What path would minimize its flying time? [5]

16
B

√ x
130

2
0

+
b = 500

50
0
C x 2 N
A a−x

Once we use the Pythagorean Theorem to find that a = 1200, all we have
to do is plug into the general equation,

(1/10)(500) 2000
x= p = .
(1/8)2 − (1/10)2 3

Therefore, the bird should fly to a point A, which is 2000/3 ≈ 666.67 meters
east of point C to minimize their time.

4.1.3 Similar Problem Examples

Example 19. An oil pipeline is built with two different kinds of tubing. From
a point P on one side of a river, the line must cross the river and then proceed
to a point Q along the bank on the other side. The tubing used in crossing the
river costs 50 percent more than the tubing which can be used on dry land.
The river is 1/2 km wide and the point Q is 5 km down river from P . To
what point R across the river must the pipe be directed so as to minimize the
total cost? [1]

Example 20. A man is in a boat 2 miles from the nearest point on the coast.
He is to go to a point Q, 3 miles down the coast and 1 mile inland. If he can
row at 2 miles per hour and walk at 4 miles per hour, toward what point on
the coast should he row in order to reach point Q in the least time? [2]

17
4.2 Pipe Corridor Problem
The Pipe Corridor problem is another frequently used problem, especially in
later textbooks. Again, we can find a generalized solution so that once we
know how to identify a Pipe Corridor problem, we can simply plug into our
generalized result.
Example 21. Find the length of the longest thin, rigid pipe that can be
carried from one 10 foot wide corridor to a similar corridor at right angles
to the first. Assume that the pipe has negligible diameter. [5]

B
x

y a

4.2.1 Generalized Solution

Let x = BD and y = DA. The problem of maximizing the length of the
pole, L can be reduced to the following constraint by similar triangles:
x y
=p .
b y − a2
2

p
Since x = by/ y 2 − a2 , we want to minimize L, but find the longest pipe,
by
L(y) = p + y.
y 2 − a2
We can take the derivative and set equal to zero to find our minimal points:
a2 b
L0 (y) = 1 − = 0,
(y 2 − a2 )(3/2)

18
(y 2 − a2 )(3/2) = a2 b,
p
y = a4/3 b2/3 + a2 .
Much like the “Triathlon Problem,” any problem with a similar set-up can
be plugged into the above equation. We will now solve the original problem
given the equation obtained.

4.2.2 Specific Solution

From the problem, we know that a = b = 10. Plugging into the general
equation, q
y = (10)4/3 (10)2/3 + 102 ,
√
y = 10 2.
√ √ √
Since x=y, we find the minimum length of the pole to be 10 2+10 2 = 20 2
feet.

4.2.3 Similar Problem Examples

Example 22. Find the length of the longest rod which can be carried horizontally
around a corner from a corridor 8 m wide to one 4 m wide. [3]

4.3 Post Problems

The Post Problems are another example of a type of optimization problem
that cannot be solved using the AM/GM, but no calculus is involved either.
After a very simple observation, all that is necessary is some basic algebra
skills to find a general solution.

Example 23. Two posts, one 8 feet high and the other 12 feet high, stand
15 ft apart. They are to be supported by wires attached to a single stake at
ground level, the wires running to the tops of the posts. Where should the
stake be placed, to use the least amount of wire? [13]

19
b
a

x (c − x)
c

4.3.1 Generalized Solution

Let a be the height of the first pole, b be the height of the second pole, and c
be the distance between the two poles. We can recognize that the angle will
remain unchanged if a straight line is drawn from (0, a) to (c, −b) as shown
below.

t b
a s

x (c − x)

t −b

We are looking for the shortest distance between these two points as we
are trying to minimize the amount of wire used. Finding the slope between
the two points, and plugging into point slope form gives:
∆y a+b a+b
= = ,
∆x 0−c c

20
a+b
y=− x + a.
c
Finding where this line intersects the x axis gives the desired result.
a+b
0=− + a,
c
ac
x= .
a+b

4.3.2 Specific Solution

From the problem, we know a = 8, b = 12 and c = 15. Plugging into the
general equation,
(8)(15)
x= = 6.
8 + 12
Thus we have found to minimize the wire used, the stake should be placed 6
feet from the 8 foot high post.

4.3.3 Similar Problem Examples

Example 24. Two saplings, 6 and 8 feet high, are planted 10 feet apart. To
prevent bending, poles the height of the trees are pounded in, then attached to
each tree, and a rope tied to the top of each pole is then fixed to the ground
(between the two trees) after being pulled taut. How close to the taller tree
will the rope be fixed if the total length of the rope is to be minimized? [1]

Example 25. Two towns, located on the same side of a straight river, agree
to construct a pumping station and filtering plant at the river’s edge, to be
used jointly to supply the towns with water. If the distances of the two towns
from the river are a and b and the distance between them is c, show that the
sum of the lengths√ of the pipe lines joining them to the pumping station is at
least as great as c2 + 4ab. [4]

5 Problems that Have Evolved

Flipping through old and new calculus textbooks, perhaps the most interesting
thing found was the evolution of some of these problems. Most early textbooks
stick to number or geometrical problems, while later ones begin making

21
more complex story problems including Triathlon, Pipe Corridor, and Post
Problems. What we find is that while later books appear to include more
complex examples, when reduced down to what they are maximizing/minimizing
and their constraints, we see that they are in fact the same problem. Below
are two examples of problems and how they evolved from their early format.

5.1 Number Problem → Box Problem

Consider the following problem:

Example 26. From a square piece of cardboard of length 20 feet, we make

an open box of maximum volume by cutting small squares out of the corners
and turning up the sides. Find the dimensions of the box. [6]

2y + x = 20

y
y

2y + x = 20

Suppose that y is the length of the corner square, x is the distance between
the y’s. The problem of maximizing the volume of the box can now be
reduced to the following:

maximize x2 y subject to 2y + x = 20.

But, if we recall from earlier the problem,

Example 27. Divide the number 20 into two parts such that the product of
one part by the square of the other part shall be a maximum. [8]

22
We are trying to maximize the product considering the following:

maximize x2 y subject to x + y = 20.

While at first glance, these problems appear very different, they are
reduced down to the same problem. From here, each problem can easily
be solved using the AM/GM inequality. Another example of a problem we
saw evolve twice from 1900 − 2016 is shown below.

5.1.1 Similar Problem Examples

Example 28. A box of maximum contents is to be made from a rectangular
piece of tin 30 inches by 14 inches; required the side of the square to be cut
out of each corner of the tin sheet. [12]

5.2 Graph Problem → Pipe Problem → Ladder Problem

Consider the following problem:

Example 29. A right triangle in the first quadrant has two coordinate axes as
sides and the hypotenuse passes through the point (10, 10). Find the vertices
of the triangle such that the length of the hypotenuse is a minimum. [8]

D (10, 10)

Without much effort, we can recognize this problem is the same problem
as the Pipe Down the Corridor problem that we solved earlier. The following
figure shows where the similarities occur.

23
b = 10

B B

D(10, 10) D

a = 10

A A
Further, we can look at yet another problem that appears different, but
is actually the same as these two problems.

Example 30. A 5-ft fence stands 4-ft from a high wall. How long is the
shortest ladder that can reach from the ground outside the fence to the wall?
[10]

4 x A

5.2.1 Similar Problem Examples

Example 31. Find the shortest length that can be drawn through a given
point (a, b) and terminate in the rectangular axes to which the point is referred.
[12]

24
6 The CS Inequality
In addition the AM/GM inequality, the Cauchy Schwarz Inequality also has
some useful applications when it comes to optimization problems. It can be
used in some situations where the max/min and constraint involve addition,
as the AM/GM inequality is not useful in problems such as these. We will
begin by proving the inequality, looking at a typical application of the CS
inequality, and then an example of an optimization problem that can be
solved using the CS inequality.

6.1 Formal Proof - CS

The Cauchy Schwarz inequality can be written in two different forms - one
in coordinate form, one in vector form. The coordinate form is as follows:

Theorem 32 (Cauchy Schwarz Inequality). Let x1 , ..., xn and y1 , ..., yn be

real numbers. Then,
q q
(x1 y1 + · · · + xn yn ) ≤ (x21 + · · · + x2n ) (y12 + · · · + yn2 ).

Equality holds if and only if either there exists a real number k such that
x1 = ky1 , x2 = ky2 ,...,xn = kyn , or there exists a real number m such that
y1 = kx1 , y2 = kx2 ,...,yn = kxn .

The vector form of the theorem follows.

Theorem 33 (Cauchy Schwarz Inequality). If X and Y are any two vectors

then
|X · Y | ≤ |X||Y |,
with equality if and only if one of the vectors is a scalar multiple of the other.

Proof. If Y = 0, then both sides are 0, so the equation holds with equality.
In this case Y = 0 = X. Now suppose that Y 6= 0 and t is any real number.
Then n
X
0≤ (xi − tyi )2
i=1
n
X n
X n
X
= x2i − 2t 2
xi y i + t yi2
i=1 i=1 i=1

25
= |X|2 − 2(X · Y )t + t2 |Y |2 .
The last expression is a second-degree polynomial p in t. From the quadratic
formula, the zeros of p are
p
(X · Y ) ± (X · Y )2 − |X|2 |Y |2
t= .
|Y |2

Hence,
(X · Y )2 ≤ |X|2 |Y |2 ,
because if not, then p would have two distinct real zeros and therefore be
negative between them, contradicting the inequality. Taking square roots
yields the inequality if Y 6= 0.
If X = tY , then |X · Y | = |X||Y | = |t||Y |2 , so equality holds. Conversely, if
equality holds, then p has the real zero t0 = (X · Y )/|Y |2 , and
n
X
(xi − t0 yi )2 = 0.
i=1

Therefore, X = t0 Y . [11]

7 Optimization Problems and the CS Inequality

Below is an example of where the CS inequality is useful in an optimization
problem. First, there is an example of a typical application of the inequality,
then we will move to an optimization problem that can be solved using the
CS inequality, but not the AM/GM inequality.

7.1 Typical Problem

The Cauchy Schwarz inequality is typically taught in multi-variable calculus.
Below is an example of the type of problem that typically shows up when
learning how to apply this inequality.

Example 34. What is the minimum value of x2 +9y 2 given that 4x+9y = 36?
[4]

26
Recognizing that 4x + 9y = 4x + 3(3y), and plugging into the inequality,
√ p
4x + 3(3y) ≤ 42 + 32 x2 + (3y)2 ,

362 ≤ 25(x2 + 9y 2 ),
362
≤ x2 + 9y 2 .
25
Therefore, the minimum value of x2 + 9y 2 subject to 4x + 9y = 36 is
approximately 51.84. From the CS inequality, we know:

x = 4t, 3y = 3t.

Solving for x and y gives:

3y 3t
= ,
x 4t
1
y = x.
4
Plugging into our constraint gives,

16y + 9y = 36,
36 144
y= , x= .
25 25

7.2 Optimization Problem

Finding applications of the CS inequality on optimization problems prove
to be much more difficult than the AM/GM. However, in certain instances,
(such as the one below), we are able to see some really concise applications
of the CS inequality.

Example 35. A length of wire 28 feet long is cut into two pieces. One piece
is bent into a 3 : 4 : 5 right triangle and the other piece is bent into a square.
Find the minimum combined area of the two shapes. [13]

27
4x 5x
y

3x y

Let y be the side length of the square, and x be the multiple of the
3 : 4 : 5 triangle. We know that the perimeter of the above figures add to
the following:
12x + 4y = 28,
3x + y = 7,
and that the combined area of the triangle and square that we are trying to
minimize is as follows:
(3x)(4x)
+ y2,
2
6x2 + y 2 .
Recognizing, √
6 √
3x + y = 6x + y,
2
then plugging into the CS inequality gives:
√ s √
2

6 √ 6
q √
2
, 1 h 6x, y ≤ +1 2 6x + y 2 ,
2 2
√

6 √
r
5p 2
, 1 h 6x, y ≤ 6x + y 2 ,
2 2
r
5p 2
3x + y ≤ 6x + y 2 ,
2
7 p
p ≤ 6x2 + y 2 ,
5/2

28
98
≤ 6x2 + y 2 .
5
Therefore, the minimum value of the area of the triangle and the square is
19.6 ft 2 .

8 Historical Bit
Looking through a hundred years of calculus problems proved to be extremely
interesting for a multitude of reasons. As was referenced in the “Problems
that Have Evolved” section, we saw a few problems that in the early 1900’s
were given in a particular way and more recently have been worded differently.
We found these problems particularly interesting, as it appears as the years
go on the authors are trying to express how calculus is useful in more
real world type problems. Also, as touched on earlier, the implementation
of other disciplines also became very evident as the years went by in the
texts. The more recent textbooks included more economic, physical, and
medical problems, we suspected as calculus became a requirement for more
disciplines. Listed below are a few examples of these problems. Another
aspect that was especially interesting to look at was the introductions of
each of these books. One of the books stated in the introduction,“In this
calculus book for young men...” Clearly, in more recent calculus books this
would not be an acceptable introduction, so seeing that mindset being phased
out after the 1913 textbook was also a neat progression to see.

8.1 Economic/Medical/Physical Problems

Example 36. In constructing the new Trump Colosseum, projected to occupy
the entire state of Rhode Island, the builder estimates the initial costs (buying
Rhode Island, etc.) as 450 times the cost of the first floor. The second floor
is projected to cost twice as much as the first floor, the third floor three times
as much as the first floor, etc. What number of floors in the building will
give the cheapest average cost per floor? [13]
Example 37. The strength of a drug is given by R(M ) where M measures
the dosage. The sensitivity of the patient’s body to the drug is the derivative.
The strength of a patients reaction to a dose of M milligrams of a certain
drug is R(M ) = c1 M 2 (c2 − M ) where c1 and c2 are positive constants. For
what value of M is the strength a maximum? [13]

29
9 Conclusion
While we got a good grasp on what the AM/GM inequality is, and when it
can be used there is still more to learn. We did many optimization problems
using this technique, however there are so many more problems out there.
Becoming familiar with more of these problems, and learning small tricks
to make the equality work would only continue to happen with exploration.
Further, a look at a few more Cauchy Schwarz inequality applications would
be an interesting direction for further research. Lastly, looking at how
optimization problems were solved before calculus existed would bring about
some unexpected results as well.

References
[1] S. Grossman Calculus Second Edition, New York, 1981.

[2] R. Larson, R.Hostetler, B. Edwards. Calculus with Analytic Geometry

Fourth Edition, D.C. Heath, 1990.

[3] M. Protter, C. Morrey. College Calculus with Analytic Geometry Third

Edition, Addison-Wesley, 1977.

[4] G. Thomas. Calculus with Analytic Geomentry Fourth Edition,

Addison-Wesley, 1968.

[5] Robert Ellis, Denny Gulick. Calculus One and Several Variables
University of Maryland, Maryland 1991.

[6] G.E.F. Sherwood, Angus E. Taylor Calculus University of California at

Los Angeles, California 1943.

[7] P. Keef, D. Guichard, R. Gordon. An Introduction to Higher Math.

Whitman College, Washington 2010.

[8] W.S. Franklin, B. MacNutt, R.L. Charles. An Elementary Treatise

South Bethlehem, Pennsylvania, 1913.

[9] J. Stewart Calculus: Early Transcendentals Seventh Edition, Cengage

Learning, 2012.

30
[10] Flanders, Prince. something 1978

[11] W. Trench. Introduction to Real Analysis Pearson Education, Trinity

University, 2015.

[12] E. Nichols. Differential and Integral Calculus with Applications D. C.

Heath & Co., Boston, 1902.

[13] R. Fraga, D. Hart, E. Herman, J. Stain, L. Welch. Calculus Problems for

a New Century The Mathematical Association of America, Washington
D.C., 1993.

31