Union College

Union | Digital Works

Honors Theses Student Work

6-2019

Geometric Constructions, Origami, and Galois

Theory
Julia Greene
Union College - Schenectady, NY

Follow this and additional works at: https://digitalworks.union.edu/theses

Part of the Mathematics Commons

Recommended Citation
Greene, Julia, "Geometric Constructions, Origami, and Galois Theory" (2019). Honors Theses. 2299.
https://digitalworks.union.edu/theses/2299

This Open Access is brought to you for free and open access by the Student Work at Union | Digital Works. It has been accepted for inclusion in Honors
Theses by an authorized administrator of Union | Digital Works. For more information, please contact [email protected].
Geometric Constructions, Origami, and Galois Theory

By

Julia Greene

*********

Submitted in partial fulfillment

of the requirements for
Honors in the Department of Mathematics

UNION COLLEGE
June, 2019
 Abstract

GREENE, JULIA Geometric Constructions, Origami, and Galois Theory. Department

of Mathematics, June 2019.

ADVISOR: George Todd

Geometric constructions using an unmarked straightedge and a compass have been stud-
ied for thousands of years. In these constructions, we can draw circles and lines starting
with any two points, and we can create new points where they intersect. An n-gon is said to
be constructible if can be constructed in a finite number of steps using these guidelines. We
begin with constructions of several n-gons, and examine the field theory behind geometric
constructions. Galois theory then provides a precise classification of which n-gons are con-
structible and which are not. Next is an exploration of origami construction, which examines
a single-fold construction axiom, and establishes the classification of origami-constructible n-
gons. For example, a heptagon is not constructible using traditional construction techniques,
but it is constructible using origami. Finally, we investigate new axioms, which might allow
for additional constructions, and examine their implications.

i
Contents

1 Introduction to Constructions 1

2 Polygon Constructions 3
2.1 Pentagon Construction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.2 Triangle Construction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.3 Square Construction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.4 Hexagon Construction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.5 Octagon Construction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.6 Heptagon Construction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

3 Field Extensions 17
3.1 The Three Problems of Antiquity . . . . . . . . . . . . . . . . . . . . . . . . 19

4 Galois Theory 22
4.1 Constructibility Criterion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

5 Origami 27
5.1 Origami Constructibility Criterion . . . . . . . . . . . . . . . . . . . . . . . . 34

6 Geometric Approach to Solving Quintic Equations 37

6.1 Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

A Background Information 44

ii
Chapter 1

Introduction to Constructions

The ancient Greeks were extremely interested in constructions that could be achieved with
an unmarked straightedge and a compass. There are several rules for constructions starting
with any two points. We will call them C1 , C2 , C3 , C4 and C5 . They are:

• C1 : A circle can be created centered at any point and through another.

• C2 : A line can be drawn through any two points.

• C3 : A point can be created where any two lines intersect.

• C4 : A point can be created where a line intersects a circle.

• C5 : A point can be created where any two circles intersect.

For example, we can construct a regular triangle starting with any two points. Given
A and B, where the distance between them is 1, we can draw a circle centered at A that
goes through B. Similarly, we can draw a circle centered at B that goes through A. Then we
have that the radius of each of these circles is 1. Now, label the point where the two circles
intersect, C. Drawing lines AB, BC, and AC yields a triangle. Each line has length 1 since
that is the radius of the circle, and so we have a regular triangle. This can be seen in Figure
1.1.
We have just seen the simple construction of a triangle, but we ultimately want to know
exactly which polygons can and cannot be constructed. We will begin exploring this question
by using a straightedge and compass to construct several different n-gons. Then we will
analytically prove that each one is actually a regular n-gon.

1
Figure 1.1: Basic Triangle Construction

2
Chapter 2

Polygon Constructions

In this chapter, we will construct a pentagon, triangle, square, hexagon and octagon using
C1 , C2 , C3 , C4 and C5 . Then we will prove that they are regular polygons, which will show
that they are constructible.

2.1 Pentagon Construction

We will start by constructing a regular pentagon using the following steps.

Start with two points, A and B, and say that the distance between them is 1. Then,
draw a circle centered at A that goes through B. This circle has radius 1. Draw another
circle centered at B that goes through A, which also has radius 1.

Figure 2.1: Pentagon Construction 1

Call the bottom point of intersection between these two circles, C. Next, draw a line
through C and A. Name the point where this line intersects the circle centered at A, D.
Then draw a line through B and D. Call the point where this line intersects the circle
centered at B, E. This can be seen in Figure 2.2.

3
 Figure 2.2: Pentagon Construction 2

Next, set the compass length as the distance between C and E.1 Now, draw a circle
centered at A with this radius. Draw a line through C and B, and label the point where it
intersects this new circle, F. Then draw an arc centered at E through F, as in Figure 2.4.
Draw a point where this arc intersects the circle centered at B. This is point H in Figure 2.5.

Figure 2.3: How to Construct Length EC

1
We can do this since it is a constructible length. We could construct it by following the same steps, but
starting with points B and C. Then A is the point of intersection between them. This can be seen in Figure
2.3.

4
 Figure 2.4: Pentagon Construction 3

Figure 2.5: Pentagon Construction 4

Now, draw an arc with the same radius centered at H. Continue this process around the
circle until there is a pentagon inscribed in the circle centered at B. This can be seen in
Figure 2.6. We will prove that this is a regular pentagon.

5
 Figure 2.6: Pentagon Construction 5

Theorem 2.1.1. This construction yeilds a regular pentagon.

Proof. To show that we have constructed a regular pentagon, we want to find the length of
one of the edges and then show that a regular pentagon would have the same edge length.
We know that EH is the same length as EF since they were constructed from the same arc,
so we will find the length of the line EF . We can do this analytically.
We want to find the distance between points E and F , so we must find the coordinates
√ each point. Let A be the origin. We know that the larger circle centered
of √ at A has radius
2 since it was constructed with a radius of EC. We know EC = 2 since ]EBC is a
right triangle and each of the legs has length 1. Thus, the circle centered at point A that
goes through F has equation,
x2 + y 2 = 2.
Then we want to find the point where this circle intersects
√ with the line that goes through
points B and C. We know that this line has slope − 3 and goes through the point (−1, 0),
thus has equation √ √
y = − 3x − 3.
Then, finding the point where the circle intersects this line, we have
 √ √ 2
x2 + − 3x − 3 = 2

x2 + 3x2 + 6x + 3 = 2
4x2 + 6x + 1 = 0.

6
 √
−6 ± 20
And applying the quadratic formula, we have that x = . But then, we must have
√ √ 8
−6 − 20 −3 − 5
that x = = , since the x value of F must be less than −1, since it is to
8 4 √ √
−3 − 5 −3 + 5
the left of B. Then we have that < −1, while > −1. Thus,
4 4
√
√ ! !
−3 − 5 √ −3 − 5 √
F = , −3 − 3 .
4 4

Similarly, we can find the coordinates of E, by finding the point where the circle

(x + 1)2 + y 2 = 1

intersects with the line through B and D, which has equation,

1 1
y = √ x+ √ .
3 3
Then we have,  2
2 1 1
(x + 1) + √ x + √ =1
3 3
1 2 1
x2 + 2x + x2 + x + = 0
3 3 3
2
4x + 8x + 1 = 0.
√
−8 ± 48
Applying the quadratic formula, we have that x = . But then, we must have that
√ √ 8
−8 − 48 −2 − 3
x = = since as before, we must have the x value of E less than -1.
8 2
Thus, √
√ ! !
−2 − 3 1 −2 − 3 1
E= ,√ +√ .
2 3 2 3
Then, applying the distance formula we can find the length of EF by finding the distance
between E and F . √
! √
!
−3 − 5 −2 − 3
Let α = − .
4 2
√ ! ! √ ! !
√ −3 − 5 √ 1 −2 − 3 1
And let β = −3 − 3 − √ +√ .
4 3 2 3
Then we have that p
d(E, F ) = α2 + β 2

7
 s √ s √
2(5 − 5) 5− 5
= = .
4 2
We rcan now check to see that if we do have a regular pentagon, each edge length would
√
5− 5 2π
be . Given a regular pentagon, each of the interior angles should be . Then,
2 5
bisecting one of these angles gives us a right triangle, which bisects the edge. Thus, we know
that the edge length is equal to 2(sin π5 ), which can be seen in Figure 2.7.

Figure 2.7: Length of Pentagon Edge

Now using the identity,

sin(5θ) = 16 sin5 θ − 20 sin3 θ + 5 sin θ,

which can be obtained using the Bromwich formula A.0.1, we must have that

5 π 3 π
    π 
sin(π) = 16 sin − 20 sin + 5 sin .
5 5 5
Now if we let x = sin( π5 ), we can solve for x to find our edge length. This gives us that

sin(π) = 0 = 16x5 − 20x3 + 5x = x(16x4 − 20x2 + 5).

And solving for x2 , we have √ √

20 ± 80 5 ± 5
x2 = = .
32 8

8
 √
5− 5 √ π 
And we know it must be since sin( π5 ) < sin( π4 ) = 22 . Hence, x = sin =
r √ r √ 8 5
5− 5 1 5− 5 π 
=2 . Then since the edge length is equal to 2 sin = 2x we have that
8 2 5
the edge length is
 
s √ s √
1 5− 5 5− 5
2 = .
2 2 2
Thus we have constructed a regular pentagon.

2.2 Triangle Construction

We can construct a triangle as in Figure 2.8, and we will show that this is a regular triangle.
We can find the edge length of the triangle analytically by finding the two points where
the circles intersect, and then finding the distance between them. We have that the circle
centered at B, the origin, has the equation x2 + y 2 = 1 and the circle centered at A has
equation (x + 1)2 + y 2 = 1. Thus we can set the two equal to find the points where they
intersect. We then have that x2 + y 2 = 1 implies that y 2 = 1 − x2 . So plugging in we have

(x + 1)2 + (1 − x2 ) = 1.

Thus, x2 + 2x + 1 + 1 − x2 = 1.
Implying, 2x + 2 = 1 ⇒ 2x = −1.
r √
3 3
And so we have that, x = − 21 . Then we can plug x in to find that y = ± =± , so
√ ! √ ! 4 2
1 3 1 3
our two points of intersection are − , and − , − . Now, since each of these
2 2 2 2
points is a vertex of the triangle, we need to determine the distance between them to find
the edge length. Using the distance formula we have,
√ ! √ !!
1 3 1 3
d − , , − ,−
2 2 2 2

√ √ !2
v
u 2
u 1 1 3 3
=t − −− + −−
2 2 2 2
√ 2
s
2 3
=
2
√
= 3.

9
 Figure 2.8: Triangle Construction

We can now check that this is a regular triangle. If so, each of the interior angles would be
2π π
. Then, we can bisect one of these angles to be by dropping a perpendicular. And thus,
3 3
π
the edge length would be equal to 2 sin . We will use the identity sin 3θ = 3 sin θ − 4 sin3 θ,
3
which we get from the Bromwich formula A.0.1. This gives us that
π 
3 π
 
0 = sin π = 3 sin − 4 sin .
3 3
π 
If we let x = sin , we have
3
0 = 3x − 4x3 = x(4x2 − 3).
√ √ !
3 π 3 √
Thus, x = = sin( π3 ). Hence, our edge length is 2 sin = 2 = 3, and so we have
2 3 2
a regular triangle, as desired.

2.3 Square Construction

We can construct a square inscribed in a circle, as in Figure 2.9, and find the edge length
analytically. Say that point A is the origin and point B has coordinates (1, 0). Then the
equation of the circle centered at A is x2 + y 2 = 1 and the circle centered at B has equation
(x − 1)2 + y 2 = 1. Then since

the distance between A and B is 1, we must have that the
1 1
point between them is 2 , 2 and the radius of the circle which goes through both A and B

10
is 12 . This then gives us the at the slope of line h is 1 and the slope of line i is −1. Then to
find the edge length, we will find where line h intersects with the circle centered at A and
where line i intersects with the circle centered at A.
We have that the equation for h is y = x and the equation for i is y = −x. Starting with
h, when we plug its equation into x2 + y 2 = 1, we have

x2 + x2 = 2x2 = 1.

And so,
1
x2 = .
2
1 1
And so this circle intersects with line h when x = √ and when x = − √ . We will use the
2 2
positive value, which is pointI in Figure
 2.9. Plugging the x-value back into y = x, we have
1 1
that point I has coordinates √ , √ . Similarly, to find the point where i intersects with
2 2
the circle centered at A, we will plug the equation y = −x into x2 + y 2 = 1. This gives us,

x2 + (−x)2 = x2 + x2 = 2x2 = 1.
1
Thus, as before we have that this circle intersects with line i when x = √ and when
2
1
x = − √ . Again, we will use the positive value, which is point J in Figure 2.9. Plugging
2  
1 1
this back in, the coordinates of point J are √ , − √ . We can now find the distance
2 2
between these two points. The distance formula yields:
s 2  2
1 1 1 1
d(I, J) = √ −√ + √ − −√
2 2 2 2
s  2
2
= 0 + √
2
2
2
=√
2
√
= 2.
√
Thus, our constructed square has an edge length of 2.

11
 Figure 2.9: Square Construction

Now we can check that this is actually a square. If so, each of the interior angles would
2π
be . As we did with the triangle, we can then bisect one of these angles by dropping
4
π
a perpendicular, which gives us an angle of . Thus, the edge length would be equal to
π  4
2 sin . Now using the identity sin(4θ) = cos θ(4 sin θ − 8 sin3 θ), which can be found
4
using the Bromwich formula A.0.1, we have
π   π   π 
0 = sin π = cos 4 sin − 8 sin3
4 4 4
√ 
2 π   π 
= 4 sin − 8 sin3 .
2 4 4
π 
So if we let x = sin then,
4
0 = −8x3 + 4x.
Thus 0 = −4x(2x2 − 1).
1 π  π  2 √
This gives us that x = √ = sin . And then 2 sin = √ = 2. Thus the edge
√ 2 4 4 2
length is 2, as desired.

12
2.4 Hexagon Construction
Now, we will construct a hexagon inscribed in a circle as in Figure 2.10. Given points A and
B, which have a distance of 1 between them, this hexagon is constructed by drawing a circle
centered at A which goes through B and then a circle centered at B which goes through
A. Then, using B and the two points of intersection between the circles, each new vertex
is created by centering a circle at a vertex and drawing it through point A. Then since the
radius of the circle centered at B is equal to 1, we have that the edge length between points
B and C, as in Figure 2.10, is 1.

Figure 2.10: Hexagon Construction

As we did with the pentagon, triangle and square, we will now check that this is a regular
2π
hexagon. Assuming it is, each interior angle is radians, and so we can bisect this angle
6
π
by dropping a perpendicular. This gives us a side length of 2 sin . We will use the identity
6
sin(6θ) = cos θ(6 sin θ − 32 sin3 θ + 32 sin5 θ), which comes from the Bromwich formula A.0.1,
π
to show that 2 sin = 1. This identity gives us that
6
π   π  π   π 
sin(π) = cos 6 sin − 32 sin3 + 32 sin5 .
6 6 6 6
Thus, π   π 
3 π 5 π
   
0 = cos 6 sin − 32 sin + 32 sin .
6 6 6 6

13
And so, π  π  π 
3 5
0 = 6 sin − 32 sin + 32 sin .
6 6 6
π 
Now, if we let x = sin , we have that
6
0 = 32x5 − 32x3 + 6x = 2x(16x4 − 16x2 + 3).
1 π  π
Factoring gives us that x = = sin . And thus, the edge length is 2 sin = 2x =
  2 6 6
1
2 = 1, as desired. Therefore, we have constructed a regular hexagon.
2

2.5 Octagon Construction

Next we are going to construct an octagon, which can be seen in Figure 2.11. As before, we
will find a side length analytically and verify that this side length indicates that it is a regular
octagon. We will find the distance between points B and N in Figure 2.11. Say that point A
is the origin, so the circle centered there has equation x2 + y 2 = 1. Then B has coordinates
(1, 0). We will now find the coordinates of point N . As with our square construction, we
know the line that goes through both A and N has a slope of 1, and therefore has equation
2 2
y = x. Then,
 setting it equal to our circle, x + y = 1, we have that point N has coordinates
1 1
√ , √ . We will now find the distance between points B and N . The distance formula
2 2
yields:

u √2 − 1 2  1 2
s v !
2  2 u
1 1
d(B, N ) = 1− √ + 0− √ =t √ + −√
2 2 2 2

u √
v !
u ( 2 − 1)2 + 1
=t
2
q√
( 2 − 1)2 + 1
= √
2
q √
(2 − 2 2 + 1) + 1
= √ .
2
p √ √ q √
4−2 2 2 2(2 − 2)
Thus, d(B, N ) = √ =
2 2
p √
√
q
2 2− 2
= = 2 − 2.
2
14
 p √
Therefore, the edge length is 2 − 2.

Figure 2.11: Octagon Construction

We will now verify that it is a regular octagon using the identity sin(8θ) = 8 cos θ(sin θ −
10 sin3 θ + 24 sin5 θ − 16 sin7 θ), which can be found with the Bromwich formula A.0.1. We
have that
π   π 
3 π 5 π 7 π
     
sin(π) = 8 cos sin − 10 sin + 24 sin − 16 sin .
8 8 8 8 8
Thus, π  π  π  π 
0 = sin − 10 sin3 + 24 sin5 − 16 sin7 .
8 8 8 8
π 
And if we let x = sin , we have
8
0 = x(1 − 10x2 + 24x4 − 16x6 )

Thus, 0 = −(2x2 − 1)(8x4 − 8x2 + 1).

p √
4 2 2− 2
And using the quadratic formula on (8x − 8x + 1) we get that x = . Thus, the
p √ ! 2
π 2− 2 p √
edge length is 2 sin = 2x = 2 = 2 − 2, as desired.
8 2

15
2.6 Heptagon Construction
We were able to construct a triangle, square, pentagon, hexagon, and octagon. However, we
did not construct a regularheptagon.
 For an n-gon to be constructible, we would be able
2π
to construct the length sin . Thus, if we were able to construct a heptagon, we would
 n 
2π
construct a length of sin . This is equal to a length of
7
v 
√ √
u s s 
1u1  3 7 + 21 −3 3 7 − 21 −3 
u
t 7− − .
2 3 2 2

We will later return to the constructibility of this length.

16
Chapter 3

Field Extensions

We now know that we can construct a regular triangle, square, pentagon, hexagon and
octagon. Given a straightedge and compass construction of an n-gon, we can determine
whether that polygon is a regular polygon, and therefore if the specific n-gon is constructible.
However, we want to be able to precisely determine which n-gons are constructible and which
n-gons are not. To do this, we need an algebraic classification of the field of constructible
numbers. We know that all constructions are completed through a finite number of steps,
and the creation of new points involves intersecting 2 lines, 2 circles or a line and a circle.
We will prove the following theorem.

Theorem 3.0.1. Let r be a constructible number. Then [Q(r) : Q] = 2k for some integer k.

Proof. Suppose r is constructible. Then we can construct r through a finite number of

steps which involve the intersections of lines and circles. We will look at each of these steps
individually.
We will first look at the case in which 2 lines are intersected. Suppose that a1 , a2 , b1 , b2 , c1
and c2 ∈ Q. Then consider the lines l and m, where l is given by

ax + by + c = 0,

and m is the line

dx + ey + f = 0.
 
ce − bf af − cd
Taking their intersection, we get the point , , which is a another point in
bd − ae bd − ae
the field Q, thus this is a degree 1 extension. Now if a, b, c, d, e and f are points of Q, we
can look at the intersection of the circles, c1 and c2 , where c1 is

x2 + y 2 + ax + by + c = 0,

and c2 is
x2 + y 2 + dx + ey + f = 0.

17
Setting c1 = c2 we have the line,

(a − d)x + (b − e)y + (c − f ) = 0,

where (a − d), (b − e) and (c − f ) are all points of Q. Therefore, we have another degree
1 extension. Lastly, we will look at the intersection of a line l and a circle c, such that
a, b, c, d, e, f ∈ Q and l is the line

ax + by + c = 0,

and c is the circle

x2 + y 2 + dx + ey + f = 0.
If we let l = c and solve for x we get

(b2 + a2 )x2 + (2ac + db2 − aeb)x + (c2 − ceb + f b2 ) = 0.

2
Let α = (bp + a2 ), β = (2ac + db2 − aeb), and γ = (c2 − ceb + f b2 ). Thus, our solution is
−β ± β 2 − 2αγ p
x= , where β 2 − 2αγ is constructible. Hence, the points of intersection
2α p
of l and c are in the field extension Q( β 2 − 2αγ) which is a degree 2 extension of Q.
Then since each one of the steps for a construction is a degree 1 or 2 extension, and a
construction must occur in a finite number of steps, multiplying the degree of each step gives
us an extension of 2k for some k. Thus, we have that [Q(r) : Q] = 2k for some integer k, as
desired.

2π



Corollary 3.0.2. If an n-gon is constructible then Q cos n
: Q = 2k for some integer
k.
 
2π
Proof. Suppose that some n-gon is constructible. Then the length cos must be con-
  n
2π
structible for the same reason that sin is constructible in our examples in Chapter 2.
n
Therefore, by Theorem 3.0.1 we must have that Q cos 2π



n
: Q = 2k for some integer k,
as desired.
We can now use the information we have on field extensions to prove whether a length is
constructible or not using its minimal polynomial as in Definition A.0.40. For example, we
r √ r √
5− 5 5− 5
saw that the length of a pentagon edge is , so we must adjoin Q with ,
2 r 2√
5− 5
which is a degree 4 extension. Then we also have that its minimal polynomial of
2
is x4 − 5x2 + 5 which is a degree 4 polynomial.

18
3.1 The Three Problems of Antiquity
There are three classic problems of antiquity involving straightedge and compass construc-
tions. The ancient Greeks knew that any angle could be bisected, but worked to determine if
an angle could be trisected. The two other problems were “doubling the cube” and “squaring
the circle.” Keeping in mind that a length, x, is constructible if [Q(x) : Q] = 2k for some k,
we can now show that all three of these problems are impossible using only a straightedge
and compass.

Example 3.1.1. Angle Trisection

An angle cannot be trisected.

π
Proof. We will show that some angle θ cannot be trisected. Let θ = . Then, if we could
3
π
trisect this angle, it would result in three equal angles of size . Thus, we would be able to
π  9 π 
construct the length, cos . We will find the minimal polynomial of a = cos to show
9 9
that a is not constructible, which implies that we cannot trisect this angle. First, we will
prove that cos(3θ) = 4 cos3 (θ) − 3 cos(θ). We know that eiθ = cos(θ) + i sin(θ), so we have,

ei3θ + e−i3θ (eiθ )3 + (e−iθ )3

cos(3θ) = =
2 2
(cos θ + i sin θ)3 + (cos θ − i sin θ)3
=
2
3
3 cosk θ(i sin θ)3−k + cosk θ(−i sin θ)3−k
X  
=
k=0
k 2
3  
X 3 i3−k + (−i3−k )
= cosk θ sin3−k θ
k=0
k 2
3    
X 3 k 3−k 1
= cos θ sin θ cos (3 − k)π
k=0
k 2

= sin θ(0) − 3 cos θ sin2 θ + cos2 θ sin θ(0) + cos3 θ

3

= cos3 θ − 3 cos θ sin2 θ = cos3 θ − 3 cos θ(1 − cos2 θ)

= cos3 θ − 3 cos θ + 3 cos3 θ
= 4 cos3 θ − 3 cos θ.
π  π 
Now, we can find the minimal polynomial for a = cos . Using the identity for cos
9 3
we have that  
3π  π   π 
cos = 4 cos3 − 3 cos .
9 9 9

19
 π 
Then if we let x = cos we have,
9
π 
cos = 4x3 − 3x.
3
π  1
And since cos = , we have
3 2
1
0 = 4x3 − 3x − .
2
π 
3
Thus, 8x − 6x − 1 has cos as a root. There is no prime in which Theorem A.0.42,
9
Eisenstein’s Criterion, allows us to verify that this is the minimal polynomial. However, we
can check that this is the minimal polynomial by shifting by −1. We can do this because
given a polynomial, f (x), if we shift by a value, a, we have f (x + a), which simply shifts the
graph horizontally. This does not affect the shape of the graph, and therefore, if f (x + a) is
irreducible, we must have that f (x) is irreducible. Now, we see that

8(x − 1)3 − 6(x − 1) − 1 = 8(x3 − 3x2 + 3x − 1) − 6(x − 1) − 1

= 8x3 − 24x2 + 18x − 3.

Which is irreducible by Eisenstein’s Criterion, Theorem A.0.42, with p = 3. Thus 8x3 −6x−1
π
must be the minimal polynomial for a = cos . But, this polynomial has degree 3, and so
h   π  i 9 π 
Q cos : Q = 3 and 3 6= 2k for any value k. Therefore, cos is not constructible
9 9
and so we cannot trisect this angle. Thus, given an arbitrary angle θ, we cannot trisect it
using the rules of straightedge and compass constructions.

Example 3.1.2. Squaring a Circle

Given a circle of radius 1, we cannot construct a square with the same area.
Proof. Suppose we have a circle with radius 1, and for contradiction assume that we can
construct a square with the same area. Since the radius of our circle is 1, the area of
the circle must be π(1)2 = π. Then we know that we can construct a square with area
π, thus the length of its edges is constructible.
√ Let s be the side length of the square. √
2
Then we know that s = π, so s = π. Then since s is constructible, we know that √ π
is constructible.√ This√ implies that we can construct π, since if we can construct π, we
can construct ( π)( π) = π. But, since π is transcendental, we know that Q(π)/Q is not
algebraic and π is not the root of any polynomial with rational coefficients. Hence, [Q(π) : Q]
is infinite. But, if π was constructible, we would have that [Q(π) : Q] = 2k for some integer
k, thus we have a contradiction, and we cannot square a circle.

Example 3.1.3. Doubling the Cube

20
 Given a cube with side length 1, we cannot construct a cube with twice the volume.
Proof. Suppose we have a cube with side length 1. Then we have that the volume of this cube
is 13 = 1. We will show that we cannot construct a cube with twice this volume. Suppose
we can. Then we want to construct a cube with volume 2. If this cube is contructible, its
side length, s must√ be constructible. Then, since the volume of the cube
√ is 2, we know that
√
s = 2, thus s = 2. We then have that the minimal polynomial of 2 is x3 − 2, since 3 2
3 3 3

is a root and it is irreducible by Eisenstein’s criterion,

√ Theorem A.0.42, using p = 2. Then,
since x − 2 has degree 3, we know that [Q( 2) : Q] = 3, and 3 6= 2k for any k. Thus, s
3 3

is not constructible, and we cannot construct a cube with twice the volume of our original
cube.

21
Chapter 4

Galois Theory

We are now able to show which extensions yield a constructible length, but we wish to
classify exactly which n-gons are constructible. We can use Galois Theory to do this.
We know that ifa polygon is constructible, then its edge length is constructible. Then
2π
the length cos is constructible, as in Corollary 3.0.2. We will use this to prove the
n
Constructibility Criterion of a Regular n-gon in Theorem 4.1.1.
We first need some definitions and theorems of Galois Theory. The following can be
found in Chapters 32 and 33 of Gallian’s book [6].

Definition 4.0.1. Roots of Unity are solutions to xn − 1 = 0. Gn = {x|xn − 1 = 0} is a

finite group under multiplication.
 √ 
1 ± −3
For example, we have that G3 = {x|x3 − 1 = 0} = 1, and G4 = {x|x4 − 1 =
2
0} = {1, −1, i, −i}.

Definition 4.0.2. A generator for Gn is a primitive nth root of unity.

Definition 4.0.3. Fix n and let ω1 , ω2 , ..., ωϕ(n) be the primitive nth roots of unity. Then
Φn (x) = (x − ω1 )(x − ω2 )...(x − ωϕ(n) ) is called the nth cyclotomic polynomial over Q.

Note that Φn has degree ϕ(n), where ϕ(n) is Euler’s phi-Function.

Theorem 4.0.4. For every positive integer n, xn − 1 = Φd (x). In particular, for p prime,
Q
d|n
p p−1 p−2 2
x − 1 = (x − 1)(x +x + ... + x + x + 1).

Theorem 4.0.5. If n ∈ N, then

(i) Φn (x) ∈ Z[x]

(ii) Φn (x) is irreducible over Z.

Theorem 4.0.6. Let ω be a primitive nth root of unity. Then, Gal(Q(ω)/Q) ∼

= U (n).

22
Definition 4.0.7. Let E be an extension field of F . Then an automorphism of E is an
isomorphism from E to E.
For example, Aut(C/R) = {id, τ }, τ (a + bi) = a − bi.
Definition 4.0.8. The Galois Group of E over F , denoted Gal(E/F ) is the set of automor-
phisms of E which take every element of F to itself. That is,

Gal(E/F ) = {φ : E → E | φ is an automorphism, φ(x) = x, ∀x ∈ F }.

2πi
Lemma 4.0.9. If n ∈ N, ω = e n , then Q(cos( 2π
n
)) ⊆ Q(ω).
Lemma 4.0.10. Consider an extension E over Q. Then for every automorphism, φ of E,
φ(x) = x for all x ∈ Q.
Proof. Let φ be an automorphism of E, and let x ∈ Q. Since x ∈ Q, we have x = m n
where m, n ∈ Z. We want to show that φ(x) = x. Since 1 is a unit, we must have that the
automorphism φ(1) = 1. Then we have that

φ(n) = φ((n)(1)) = φ (1 + 1 + ... + 1) = φ(1) + φ(1) + ... + φ(1) = n(φ(1)) = (n)(1) = n.

| {z } | {z }
n times n times

Similarly, φ(m) = m. We also have that

     
1 1 1
φ(1) = φ n = φ(n)φ = nφ = 1.
n n n
 
1 1
So we must have that φ = . Now, combining our results, we have that
n n
m      
1 1 1 m
φ(x) = φ =φ m = φ(m)φ =m = = x.
n n n n n
Hence, φ(x) = x, as desired.
√
Example 4.0.11. Let Gal(Q( 2)/Q) = G.
Then, by Lemma 4 we have that φ(x) = x for all x ∈ Q and for all φ ∈ G. Thus
√ √
φ(a + b 2) = φ(a) + φ(b)φ( 2)
√
= a + bφ( 2).
√
Thus, φ(x) is completely determined by φ( 2). Then we see that,
√ √ √
2 = φ(2) = φ( 2 2) = (φ( 2))2 .
√ √
So, φ( 2) = ± 2√and we have√that G has 2 elements, the identity mapping and the mapping
which sends a + b 2 to a − b 2. Hence, |G| = 2, and G ∼= Z/2Z.

23
Theorem 4.0.12 (The Fundamental Theorem of Galois Theory). Let F be Q or an extension
of Q. If E/F is the splitting field for some polynomial in F [x] (E/F is a Galois extension),
then
ϕ : {F ⊆ K ⊆ E} 7→ {H|H ≤ Gal(E/F )}
K 7−→ Gal(E/F )
is one-to-one. We also have the following four facts:

1. [E : K] = |Gal(E/K)| and [K : F ] = |Gal(E/F )|/|Gal(E/K)|

2. If K is the splitting field of some polynomial in F [x] then,

Gal(E/F )
Gal(K/F ) ∼
=
Gal(E/K)

3. K = EGal(E/K)

4. If H ≤ Gal(E/F ) then H = Gal(E/EH )

We now have the information we need to prove our constructability criterion. To char-
acterize constructible n-gons, we will prove the following:

4.1 Constructibility Criterion

Theorem 4.1.1 (Constructability Criterion). A regular n-gon is constructible if and only if
n = 2k p1 p2 ...pt where k ≥ 0 and each pi is a prime of the form pi = 2mi + 1 for some mi .

Proof. (⇒) Suppose that a regular n-gon is constructible. We will show that n = 2k p1 p2 ...pt
where k ≥ 0 and pi is a prime such that pi = 2mi + 1. Since our n-gon is constructible,


by
2π 2π


 k
Corollary 3.0.2, Q cos n : Q = 2 . Now by



that Q cos n ⊆
Lemma 4.0.9, we know
Q(ω) where Q(ω) is an extension of Q cos 2π n
and Q cos 2π
n
is an extension of Q.
Then, since Φn (x), which has degree ϕ(n), is the minimal polynomial of ω we have that
[Q(ω) : Q] = ϕ(n). Then, by field theory we know that
        
2π 2π
[Q(ω) : Q] = Q(ω) : Q cos · Q cos :Q .
n n

Then by the Fundamental Theory of Galois theory, which is Theorem 4.0.12, we have that
        
 2π  · Q cos 2π

ϕ(n) = Gal(Q(ω))/Q cos /Q .
n  n

Or,     
2π ϕ(n)
Q cos /Q =  2π


 .
n Gal(Q(ω))/Q cos n


24
We also know that Gal(Q(ω))/Q cos 2π 2π





n
≤ Gal(Q(ω))/Q). Now, let H = Gal(Q(ω))/Q cos n
and let G = Gal(Q(ω))/Q). Then we have that H ≤ G. Now by the definition of the
Galois group,


Definition know that for all σ ∈ H, σ(ω) = ω k for some k, and

4.0.8, we 2πi
σ cos n = cos n . But, ω = e n = cos( 2π
2π 2π
n
) + i sin( 2π
n
). Thus,
    
2π 2π
σ(ω) = σ cos + i sin
n n
     
2π 2π
= σ cos + σ(i)σ sin .
n n
We also have that σ(ω) = ω k = cos 2πk + i sin 2πk . Hence, cos 2π = cos 2πk





n n n n
, which
occurs when k = 1 or k = n − 1, and so we must have |H| = 2. Then, we have that
    
2π ϕ(n) ϕ(n)
Q cos /Q = = .
n |H| 2
And by our supposition, we know that Q cos 2π



n
/Q = 2k , so we have that ϕ(n) = 2k−1
hence ϕ(n) must be a power of 2. Now suppose that pn1 1 pn2 2 ...pnt t is the prime factorization of
n. Then as shown on page 163 of Tattersall’s book [10] Euler’s phi-function is multiplicative.
Thus,
2k = ϕ(n) = ϕ(pn1 1 )ϕ(pn2 2 )...ϕ(pnt t )
= ((p1 − 1)(pn1 1 −1 ))((p2 − 1)(pn2 2 −1 ))...((pt − 1)(pnt t −1 )).
So, we must have that for each i = 1, ..., t either pi = 2 or, ni = 1 and pi − 1 = 2mi , which
implies that pi = 2mi + 1, for some mi , as desired.
(⇐) Suppose that n = 2k p1 p2 ...pt where k ≥ 0 and each pi is a prime of the form
pi = 2mi + 1 for some mi . We want to show that an n-gon is constructible by showing that
    
2π
Q cos /Q = 2k for some k.
n
We know that
ϕ(n) = [Q(ω) : Q] = |Gal(Q(ω)/Q|.
Then, by the Fundamental Theorem of Galois Theory, we know that
        
2π  2π 
Q cos /Q = |Gal(Q(ω)/Q| / Gal(Q(ω)/Q cos
 .
n n 

Then we have that,

        
2π  2π 
Q cos /Q = ϕ(n)/ Gal(Q(ω)/Q cos .
n n 

Then we know from (⇒) that

   
Gal(Q(ω)/Q cos 2π
 
 = 2.
 n 

25
So,           
2π  2π 
Q cos /Q = Gal Q cos /Q  = ϕ(n)/2.
n n
Then we know that
ϕ(n) = ϕ(2k p1 p2 ...pt )
= ϕ(2k )ϕ(p1 )ϕ(p2 )...ϕ(pt )
= 2k−1 (p11−1 (p1 − 1))...(p1−1
t (pt − 1)).
Thus we have that
ϕ(n)/2 = 2k−2 (p1 − 1)...(pt − 1).
Where pi = 2mi + 1. So we must have

ϕ(n)/2 = 2k−2 (2m1 )(2m2 )...(2mt )

= 2k for some k.
Hence,     
2π
Q cos /Q = 2k for some k.
n
And so we have that this n-gon is constructible.
We can now exactly characterize constructible polygons. We know that an n-gon is
constructible if and only if n = 2k p1 p2 ...pt where k ≥ 0 and pi is a prime such that pi =
2mi + 1. We have already seen that a triangle, square, pentagon, hexagon, and octagon are
all constructible, but we can now say that a 17-gon is constructible as well as a 257-gon. In
addition, even though we may not necessarily know how to construct a given n-gon with a
straightedge and compass, we can determine whether or not it is possible.

26
Chapter 5

Origami

We will now use origami, or paper folding, to study some constructions that are not possible
with only a straightedge and compass. In this section, the previous construction rules apply
but we will add a new rule which allows us to create lines by folding the paper. Cox [3] gives
us the following definition of a line constructed by origami.

Definition 5.0.1. Given two points a1 6= a2 not lying on lines l1 6= l2 , we can draw a new
line l, called an origami line which reflects point a1 to a new point b1 lying on l1 and reflects
a2 to a point b2 on l2 . We will call this axiom, O6 .

In this section, the rules of construction are as follows:

Starting with any two points,

• C1 : A circle can be created centered at any point and through another.

• C2 : A line can be drawn through any two points.

• C3 : A point can be created where any two lines intersect.

• C4 : A point can be created where a line intersects a circle.

• C5 : A point can be created where any two circles intersect.

• O6 : An origami line can be drawn as outlined in Definition 5.0.1.

Example 5.0.2. We can trisect an angle using origami.

27
 We begin with a square piece of origami paper, created from lines j, k, l and m in Figure
5.1. Call the point of intersection between m and j point P1 . Then, draw a line through
P1 , which is line n in our figures, that creates an arbitrary angle that we will call θ. This is
angle ]AP1 D in Figure 5.1. We will trisect this angle.

Figure 5.1: Origami Trisect 1

We will now make an origami line as in Definition 5.0.1 by reflecting P1 to a point called
P2 on line m and point A to a new point, called F , on line k. Our origami line p is parallel
to line j since j is perpendicular to both m and k. This line is the dotted line in in Figure
5.2.

Figure 5.2: Origami Trisect 2

Now draw a line through points P2 and F as in Figure 5.3.

28
 Figure 5.3: Origami Trisect 3

We will now fold another line. Fold the paper to reflect point P1 to a new point on line
p, called Q1 and P2 to a new point on line n, called Q2 . This can be seen as the dotted line
r in Figure 5.4.

Figure 5.4: Origami Trisect 4

Lastly, we will draw a line through points P1 and Q1 . It is easy to check that the angle
θ
created between lines j and this new line, s, is by folding the paper twice. We can also
3
θ
check that the angle is using the properties of similar triangles. Thus we have trisected
3
our angle, and it can seen in Figure 5.5.

29
 Figure 5.5: Origami Trisect 5

As we saw in Example 5.0.2, the rules of origami allow constructions that were previously
impossible. We want to think about what allows us to create new lines by folding and what
implications this leads to in terms of constructibility.
The lines we are creating can actually be represented as simultaneous tangents to parabo-
las. Start with a point P called the focus, and a line l, the directrix, as in Figure 5.6

Figure 5.6: Simultaneous Tangents 1

Then, by definition, a parabola is all the points which are equidistance from our focus
P and the directrix l. For example, we can take any point on the parabola - we use point
A in Figure 5.7. Then, we have that the distance from point P to point A is equal to the

30
distance between point A and line l if we drop a perpendicular line from A. This is true for
any point on our parabola.

Figure 5.7: Simultaneous Tangents 2

Now consider the tangent line to the parabola at point A as in Figure 5.8. This line can
be thought of as an origami line which reflects point P to point D on l.

Figure 5.8: Simultaneous Tangents 3

For example, if we return to the example of trisecting an angle, we can look at the origami
line in terms of simultaneous tangents. In Figure 5.9, we can draw a parabola with point P1
as the focus and line p as the directrix. Then we have that the folded line r is tangent at a
point which is equidistant to points P1 and Q1 . Similarly, if we draw a parabola with P2 as
the focus and line n as the directrix, we have that the origami line r is tangent at a point
equidistant from P2 and Q2 .

31
 Figure 5.9: Simultaneous Tangents 4

Let O be the field of numbers which are constructible with origami. We want to be able
to exactly describe all origami-constructible numbers. We continue to use the traditional
construction rules in addition to origami, so we can still intersect lines and circles to get
new points, which are in degree 1 or degree 2 field extensions. We will now look at the new
points created with origami lines.
As outlined by Cox, [3] we will look at the simultaneous tangents of two parabolas
 2
1 1
y − a = 2bx and y = x2 .
2 2
 2
1
Starting with y − a = 2bx we can use its derivative to find the slope of its tangent,
2
which is
b
m= .
y − 12 a
Then
  a point (x1 , y1 ) on the parabola in terms of m, and we have that (x1 , y1 ) =
we can define
b b a 1
2
, + . Now looking at y = x2 , we can find its derivative to show that the slope
2m m 2 2
of its tangent line is
m = x.
2
m
Thus we have that (x2 , y2 ) = (m, ). Now we will look at the slope formula,
2
y2 − y1
m= .
x2 − x1

32
Using our points (x1 , y1 ) and (x2 , y2 ), we have

m2
 
b a
− +
2 m 2 m4 − 2bm − am2
m= = .
b 2m3 − b
m−
2m2
Thus m is the solution to the cubic equation,

m3 + am + b = 0.

This gives us the fact that we can now solve any cubic m3 + am + b = 0 where a and b are
constructible using the simultaneous tangent of two parabolas.
Then Lee [8] explains how this implies that m is a constructible distance. We can con-
struct m by dropping a perpendicular line which is a distance 1 from a constructible point
on the simultaneous tangent line. This can be seen in Figure 5.10.

Figure 5.10: Constructing the Distance m

Then since m is the solution of an irreducible degree 3 polynomial, we must have that
the degree of the extension from K to K(m) is a degree 3 extension.
Let x ∈ O. Then we can construct the length x in a finite number of steps using
C1 , C2 , C3 , C4 , C5 and O6 . Then since each of the steps for a construction is a degree 1,2, or
3 extension, multiplying the degree of each step gives us an extension of 2k 3j for some k.
Thus, we have that [Q(x) : Q] = 2k 3j for some integers k and j.
We can now prove that a heptagon is constructible using origami.

Theorem 5.0.3. A heptagon is constructible with origami.

Proof. As we did with the polygons in Chapter 2, in order to show that a heptagon is
constructible, we need to adjoin an element to Q and find the degree of the field extension.

33
We know by Section 2.6, that if a heptagon is constructible then the length
v 
√ √
u s s 
 
2π 1u1 3 7 + 21 −3 3 7 − 21 −3 
u
sin = t 7 − −
7 2 3 2 2

is constructible. Thus, we want to determine if this

 length
 is constructible using origami. We
2π
can verify that 64x6 − 112x4 + 56x2 − 7 has sin as a root and using Theorem A.0.42,
7  
2π
Eisenstein’s Criterion, with p = 7 we see that this is the minimal polynomial of sin .
7
This polynomial has degree 6 = 21 31 . Hence, a heptagon is constructible using origami.
We now want to characterize the n-gons that are constructible using origami.

5.1 Origami Constructibility Criterion

Theorem 5.1.1 (Origami Constructibility Criterion). A regular n-gon is constructible with
origami if and only if n = 2a 3b p1 p2 ...pt where a, b ≥ 0 and each pi is a prime of the form
pi = 2mi 3ni + 1 for some mi and ni .
Proof. (⇒) Suppose that a regular n-gon is constructible by origami. We will show that
n = 2a 3b p1 p2 ...pt where a, b ≥ 0 and pi is a prime


such
 that pi = 2mi 3ni + 1. Since our
2π k j
n-gon is constructible, we know that Q cos n : Q = 2 3 . We

know by Lemma

4.0.9
2π 2π 2π



that Q cos n ⊆ Q(ω) where Q(ω) is an extension of Q cos n and Q cos n is an
extension of Q. Then, since Φn (x), which has degree ϕ(n), is the minimal polynomial of ω
we have that [Q(ω) : Q] = ϕ(n). Now, by field theory we have that
        
2π 2π
[Q(ω) : Q] = Q(ω) : Q cos · Q cos :Q
n n
Then by the Fundamental Theory of Galois theory, or Theorem 4.0.12, we have that
        
 2π  · Q cos 2π

ϕ(n) = Gal(Q(ω))/Q cos /Q ,
n  n
or     
2π ϕ(n)
Q cos /Q = 

 .
n Gal(Q(ω))/Q cos 2π n

We also know that Gal(Q(ω))/Q cos 2π 2π





n
≤ Gal(Q(ω))/Q). Now, let H = Gal(Q(ω))/Q cos n
and let G = Gal(Q(ω))/Q). Then we have that H ≤ G. Now by the definition of the
Galois group, Definition 4.0.8, we know that for all σ ∈ H, σ(ω) = ω k for some k, and
2πi
σ cos 2π = cos 2π . But, ω = e n = cos( 2π ) + i sin( 2π




n n n n
). Thus,
    
2π 2π
σ(ω) = σ cos + i sin
n n

34
      
2π 2π
= σ cos + σ(i)σ sin .
n n
We also have that σ(ω) = ω k = cos 2πk + i sin 2πk . Hence, cos 2π = cos 2πk





n n n n
, which
occurs when k = 1 or k = n − 1. Thus, |H| = 2. Then, we have that
    
2π ϕ(n) ϕ(n)
Q cos /Q = = .
n |H| 2

And by our supposition, we know that Q cos 2π




n
/Q = 2k 3j , so we have that ϕ(n) = 2l 3j ,
n1 n2 nr
where l = k − 1. Let q1 q2 ...qr be the prime factorization of n. Then since Euler’s phi-
function is multiplicative [10] we have that

2l 3j = φ(n) = ((q1 − 1)(q1n1 −1 ))((q2 − 1)(q2n2 −1 ))...((qr − 1)(qrnr −1 )).

Thus, we must have that qi = 2 or 3 or qi − 1 = 2mi 3ni for some mi and ni . Thus, we must
have that n = 2a 3b p1 p2 ...pt where pi = 2mi 3ni + 1.
(⇐) Suppose that n = 2a 3b p1 p2 ...pt where a, b ≥ 0 and pi = 2mi 3ni + 1. We want to show
that an n-gon is constructible by showing that
    
2π
Q cos /Q = 2j 3k for some j and k.
n
We know that
ϕ(n) = [Q(ω) : Q] = |Gal(Q(ω)/Q|.
Then, by the Fundamental Theorem of Galois Theory, we know that
        
2π  2π 
Q cos /Q = |Gal(Q(ω)/Q| / Gal(Q(ω)/Q cos
 .
n n 

Then we have that,

        
2π  2π 
Q cos /Q = ϕ(n)/ Gal(Q(ω)/Q cos .
n n 

Then we know from (⇒) that

   
Gal(Q(ω)/Q cos 2π
 
 = 2.
 n 

So,           
2π  2π 
Q cos /Q = Gal Q cos /Q  = ϕ(n)/2.
n n
Then we know that
ϕ(n) = ϕ(2a 3b p1 p2 ...pt )
= ϕ(2a )ϕ(3b )ϕ(p1 )ϕ(p2 )...ϕ(pt )

35
 = 2a−1 3b−1 (p1−1 1−1
1 (p1 − 1))...(pt (pt − 1)).
Then we have that
ϕ(n)/2 = 2a−2 3b−1 (p1 − 1)(p2 − 1)...(pt − 1).
Where pi = 2mi 3ni + 1. So we have that

ϕ(n)/2 = 2a−2 3b−1 (2m1 3n1 )(2m2 3n2 )...(2mt 3nt )

= 2j 3k for some j and k.

Thus, we have that
    
2π
Q cos /Q = 2j 3k for some j and k.
n

And so we have that this n-gon is constructible by origami.

We now know exactly which n-gons are constructible using origami. We have shown that
a heptagon is constructible, but we also know that a 54-gon and a 22-gon are constructible
with origami. We saw that with our traditional construction techniques that the following
n-gons are constructible:

3, 4, 5, 6, 8, 10, 12, 15, 16, 17....

However, we now have that the following are constructible with origami:

3, 4, 5, 6, 7, 8, 9, 10, 12, 15, 16, 17, 18, ....

We can also return to the three problems of antiquity. We saw that we can trisect an angle
with origami in Example√5.0.2, but we can also say that doubling a cube is possible with the
addition of O6 since [Q( 3 2) : Q] = 3. Origami expanded our construction possibilities, and
we now want to look at additional construction techniques.

36
Chapter 6

Geometric Approach to Solving Quintic

Equations

We saw in Chapter 5 that an n-gon is constructible by origami if and only if n = 2k 3j p1 p2 ...pt

where j, k ≥ 0 and each pi is a prime number of the form pi = 2l 3m + 1. This is because our
new origami construction axiom allowed us to construct solutions to cubic equations. We
want to extend this further by creating a new construction axiom which would allow us to
solve quintic equations. Suppose such an axiom exists. We will call it J7 . Now suppose that
J is the field of numbers that are constructible using C1 , C2 , C3 , C4 , C5 , O6 and J7 . Since J7
allows us to solve quintic equations, it would give us the following results:
Proposition 6.0.1. If a length x ∈ J then [Q(x) : Q] = 2a 3b 5c for some integers a, b and c.
We have already seen that C1 , C2 , C3 , C4 , C5 and O6 give us field extensions of degree 2
and 3. Now with J7 , we have the addition of degree 5 extensions, and therefore we have that
[Q(x) : Q] = 2a 3b 5c .
Proposition 6.0.2. An n-gon is constructible with C1 , C2 , C3 , C4 , C5 , O6 and J7 if and only
if n = 2k 3j 5l p1 p2 ...pt where j, k, l ≥ 0 and pi is a prime of the form pi = 2ai 3bi 5ci + 1.
Proof. (⇒) Suppose that a regular n-gon is constructible by origami. We will show that
n = 2k 3j 5l p1 p2 ...pt where  l ≥ d0 eand


k, j, pi = 2ai 3bi 5ci +1. Since our n-gon is constructible, we
2π 2π
f



know that Q cos n : Q = 2 3 5 .

We know by Lemma 4.0.9 that Q cos n
⊆ Q(ω)
where Q(ω) is an extension of Q cos 2π 2π



n
and Q cos n
is an extension of Q. Then, since
Φn (x), which has degree ϕ(n), is the minimal polynomial of ω we have that [Q(ω) : Q] = ϕ(n).
Now, by field theory we know
        
2π 2π
[Q(ω) : Q] = Q(ω) : Q cos · Q cos :Q .
n n
Then by the Fundamental Theory


  of Galois theory,
 Theorem 4.0.12, we must have that
2π 2π



ϕ(n) = Gal(Q(ω))/Q cos n
  · Q cos n /Q , or
    
2π ϕ(n)
Q cos /Q = 

 .
n Gal(Q(ω))/Q cos 2π
n


37
We also know that Gal(Q(ω))/Q cos 2π 2π





n
≤ Gal(Q(ω))/Q). Now, let H = Gal(Q(ω))/Q cos n
and let G = Gal(Q(ω))/Q). Then we have that H ≤ G. Now by the definition of the
Galois group,


Definition know that for all σ ∈ H, σ(ω) = ω k for some k, and

4.0.8, we 2πi
σ cos n = cos n . But, ω = e n = cos( 2π
2π 2π
n
) + i sin( 2πn
). Thus,
    
2π 2π
σ(ω) = σ cos + i sin
n n
     
2π 2π
= σ cos + σ(i)σ sin .
n n
We also have that σ(ω) = ω k = cos 2πk 2πk 2π 2πk





n
+ i sin n
. Hence, cos n
= cos n
, which
occurs when k = 1 or k = n − 1. Thus, |H| = 2. Then, we have that
    
2π ϕ(n) ϕ(n)
Q cos /Q = = .
n |H| 2

And by our supposition, we know that Q cos 2π




n
/Q = 2d 3e 5f , so we have that ϕ(n) =
m e f n1 n2 qr
2 3 5 , where m = d − 1. Let q1 q2 ...qr be the prime factorization of n. Thus

2m 3e 5f = ϕ(n) = ((q1 − 1)(q1n1 −1 ))((q2 − 1)(q2n2 −1 ))...((qr − 1)(qrnr −1 )).

Therefore, we must have qi = 2, 3 or 5, or qi − 1 = 2ai 3bi 5ci . Thus, we must have that
n = 2k 3j 5l p1 p2 ...pt where pi is a prime of the form pi = 2a 3b 5c + 1.
(⇐) Suppose that n = 2k 3j 5l p1 p2 ...pt where j, k, l ≥ 0 andpi is a prime such that pi =
ai bi ci
2 3 5 + 1. We want to show that an n-gon is constructible by showing that
    
2π
Q cos /Q = 2d 3e 5f for some d,e and f.
n
We know that
ϕ(n) = [Q(ω) : Q] = |Gal(Q(ω)/Q|.
Then, by the Fundamental Theorem of Galois Theory, Theorem 4.0.12, we know that
        
2π  2π 
Q cos /Q = |Gal(Q(ω)/Q| / Gal(Q(ω)/Q cos
 .
n n 

Thus we have that,

        
2π  2π 
Q cos /Q = ϕ(n)/ Gal(Q(ω)/Q cos .
n n 

Then we know from (⇒) that

   
Gal(Q(ω)/Q cos 2π
 
 = 2.
 n 

38
So,           
2π  2π 
Q cos /Q = Gal Q cos /Q  = ϕ(n)/2.
n n
Then we know that
ϕ(n) = ϕ(2k 3j 5l p1 p2 ...pt )
= ϕ(2k )ϕ(3j )ϕ(5l )ϕ(p1 )ϕ(p2 )...ϕ(pt )
= 2k−1 3j−1 5l−1 (p11−1 (p1 − 1))...(p1−1
t (pt − 1)).
Then we have that

ϕ(n)/2 = 2k−2 3j−1 5l−1 (p1 − 1)(p2 − 1)...(pt − 1).

Where pi = 2ai 3bi 5ci + 1. So we have that

ϕ(n)/2 = 2k−2 3j−1 5l−1 (2a1 3b1 5c1 )...(2at 3bt 5ct )

= 2d 3e 5f for some d,e and f.

Hence,     
2π
Q cos /Q = 2d 3e 5f for some d,e and f.
n
And so we have shown that this n-gon is constructible using our original construction tech-
niques, origami, and the addition of J7 .
We now want to find some axiom which would allow us to solve quintic equations. In
Lucero’s article [9] a 2-fold axiom is described, in which there are 2 points, A and B and 3
lines, l,m, and n. We can then simultaneously fold 2 lines, α and β. The line α places A
onto l and β places B onto m while aligning n and α. Alperin and Lang [2] also describe
this 2-fold axiom. This axiom can be used to solve specific quintic equations, however, in
general a degree n polynomial can be solved with n-2 simultaneous folds.
We will now look at a new axiom which starts with 3 points and 3 lines. We define J7 as
follows:

Definition 6.0.3. Consider 3 distinct points, A, B and C, and three distinct lines, l, m and
n such that A is not on l, B is not on m and C is not on n. Then a new line i, which we
will call a J7 line, can be drawn which reflects A to a point on l, B to a point on m and C
to a point on n. This can be seen in Figures 6.1 and 6.2.

39
 Figure 6.1: J7 Line 1

Figure 6.2: J7 Line 2

Similar to an origami line, we can then see that this line is a simultaneous tangent to
three different parabolas, as in Figure 6.3. Each of these parabolas is defined by its focus and
directrix and the simultaneous tangent reflects each focus to a new point on the directrix.

40
 Figure 6.3: J7 Line 3

As an example, we will now look at the simultaneous tangent of the three parabolas:
 2
1 2 1 2 1
y = x , x = y , and y − a = 2bx.
2 2 2
1
Starting with y = x2 , we can find its derivative to show that the slope of its tangent line is
2
m = x.

Thus if we want to define a point (x1 , y1 ) in terms of this parabola and m, we have that
m2 1
x1 = m and y1 = . We can follow the same process for x = y 2 . We have that the slope
2 2
of its tangent is
1
m= .
y
   2
1 1 1
Thus, (x2 , y2 ) = , . Lastly, we will repeat the same process for y − a = 2bx.
2m2 m 2
And we find that the slope of its tangent is
b
m= .
y − 12 a
 
b b a
Then we have that (x3 , y3 ) = , + .
2m2 m 2

41
 Now, to find the simultaneous tangent to all three parabolas we want to find the solution
for m when
y2 − y1 y3 − y1
= .
x2 − x1 x3 − x1
So if we plug in our values for (x1 , y1 ), (x2 , y2 ) and (x3 , y3 ) we have

1 m2 b a m2
− + −
m 2 = m 2 2 .
1 b
−m −m
2m2 2m2
Then simplifying we have that

2m − m4 2bm + am2 − m4
= .
1 − 2m3 b − 2m3
It then follows that
2am5 + (3b − 3)m4 − am2 = 0.
Thus, m is the solution to a quintic equation.

6.1 Future Work

We have a shown a specific example of how the J7 axiom allows us to solve a quintic equation.
However, using origami we know that we can solve for any cubic equation, m3 + am + b = 0
when a and b are constructible. We want to a similar conclusion using the J7 fold. We want
to be able to solve any quintic, m5 + am + b = 0 where a and b are constructible. We could
then look further to find a geometric solution to 7th degree polynomials, as well as looking
at a simultaneous tangent to n parabolas where n ≥ 3.

42
Bibliography

[1] Roger C Alperin. A mathematical theory of origami constructions and numbers. New
York J. Math, 6(119):133, 2000.

[2] Roger C Alperin and Robert J Lang. One-, two-, and multi-fold origami axioms.
Origami, 4:371–393, 2009.

[3] D.A. Cox. Galois Theory. Pure and Applied Mathematics: A Wiley Series of Texts,
Monographs and Tracts. Wiley, 2012.

[4] B. Carter Edwards and Jerry Shurman. Folding quartic roots. Mathematics Magazine,
74(1):19–25, 2001.

[5] Euclid, T.L. Heath, and D. Densmore. Euclid’s Elements: all thirteen books complete
in one volume : the Thomas L. Heath translation. Green Lion Press, 2002.

[6] J. Gallian. Contemporary Abstract Algebra. Cengage Learning, 2016.

[7] Thomas C Hull. Solving cubics with creases: he ork of beloch and lill. The American
Mathematical Monthly, 118(4):307–315, 2011.

[8] Hwa Young Lee. Origami-constructible numbers. PhD thesis, University of Georgia,
2017.

[9] Jorge C Lucero. On the elementary single-fold operations of origami: reflections and
incidence constraints on the plane. arXiv preprint arXiv:1610.09923, 2016.

[10] James J Tattersall. Elementary number theory in nine chapters. Cambridge University
Press, 2005.

[11] E.W. Weisstein. CRC Concise Encyclopedia of Mathematics. CRC Press, 2002.

43
Appendix A

Background Information

Bromwich Multiple Angle Formula

In his book, Eric Weisstein [11] outlined Bromwich’s multiple angle formula as follows:
Theorem A.0.1. For any given n value,
n(n2 − 12 )x3 n(n2 − 12 )(n2 − 32 )x5 )

nx −
 + − ... for n odd
sin(na) =  3! 2 2 3 2
5! 2 2 2 5

(n − 2 )x (n − 2 )(n − 4 )x
n cos(a) x −
 + − ... for n even.
3! 5!

where x = sin(a).

Rings

The following can be found in Chapter 12, “Introduction to Rings,” of Gallian’s book [6].
Definition A.0.2. A ring R is a set with two binary operations, addition (denoted by a + b)
and multiplication (denoted by ab), such that for all a, b, c in R:
1. a + b = b + a.

2. (a + b) + c = a + (b + c).

3. There is an additive identity 0. That is, there is an element 0 in R such that a + 0 = a

for all a in R.

4. There is an element −a in R such that a + (−a) = 0.

5. a(bc) = (ab)c.

6. a(b + c) = ab + ac and (b + c)a = ba + ca.

44
Example A.0.3. Let G and H be additive abelian groups. Then the set Hom(G, H) = {ϕ :
G → H|ϕ is a homomorphism } is a ring under the operations (ϕ + ψ)(x) = ϕ(x) + ψ(x)
for all x ∈ G and (ϕψ)(x) = (ϕ ◦ ψ)(x) for all x ∈ G.
Definition A.0.4. If S ⊆ R, R is a ring, and S is a ring with the same operations as R,
then S is called a subring.
Example A.0.5. The set of Gaussian integers Z[i] = { a + bi|a, b ∈ Z} is a subring of the
complex numbers C.
√ √ √
Example A.0.6. The set Q( d), d ∈ Z where Q( d) = {x + y d|x, y ∈ Q} is either a
subring of R if d ≥ 0 or a subring of C if d < 0.

Integral Domains

The following is from Chapter 13, “Integral Domains,” in Gallian’s book [6].
Definition A.0.7. A commutative ring with identity and no zero divisors is called an integral
domain.
Example A.0.8. R ⊕ R is not an integral domain since it has zero divisors. For example,

(1, 0) · (0, 1) = (0, 0).

Theorem A.0.9. Suppose D is an integral domain. Then if a, b, c ∈ D and if a 6= 0, then

ad = ac ⇒ d = c.
Definition A.0.10. If a ∈ R, R is any ring, and a has a multiplicative inverse, then a is
called a unit.
Example A.0.11. In Q, every non-zero element is a unit.
Definition A.0.12. A field is a commutative ring with identity such that every non-zero
element is a unit.
Theorem A.0.13. Every finite integral domain is a field.

Ideals and Factor Rings

The following can be found in Chapter 14, “Ideals and Factor Rings,” in Gallian’s book [6].
Definition A.0.14. If A ⊆ R is a subring, then A is called an ideal if for all r ∈ R, and
for all a ∈ A, ra ∈ A and ar ∈ A.

45
Theorem A.0.15. If A ⊆ R, then A is an ideal if

1. a, b ∈ A ⇒ a − b ∈ A

2. a ∈ A and r ∈ R ⇒ ra ∈ A and ar ∈ A.

Example A.0.16. For any ring R, R and {0} are ideals.

Example A.0.17. Let a ∈ R and consider hai = {ra|r ∈ R}. This set is an ideal called the
principal ideal generated by a.

Theorem A.0.18. Let A be a subring of R. Then A is an ideal if and only if {r + A|r ∈ R}

is a ring under
(s + A) + (t + A) = (s + t + A)
(s + A)(t + A) = st + A.

Notation. If I is an ideal of R, we write R/I = {r + I|r ∈ R}.

Definition A.0.19. Let A be an ideal of a commutative ring, R, and let A 6= R. Then if

ab ∈ A ⇒ a ∈ A or b ∈ A then we say that A is a prime ideal.

Definition A.0.20. If A is a proper ideal of R such that whenever B is an ideal and A ⊆

B ⊆ R, implies that B = A or B = R then we call A a maximal ideal.

Theorem A.0.21. Suppose R is a commutative ring with identity. Then R/A is an integral
domain if and only if A is prime.

Theorem A.0.22. Suppose R is a commutative integral domain. Then R/A is a field if and
only if A is maximal.

Ring Homomorphisms

The infromation on Ring Homomorphisms can be found in Chapter 15 of Gallian’s book [6].

Definition A.0.23. Let R,S be rings, and suppose that ϕ : R → S is a map such that

ϕ(a + b) = ϕ(a) + ϕ(b)

ϕ(ab) = ϕ(a)ϕ(b).
Then we call ϕ a ring homomorphism.

Example A.0.24. ϕ : C → R[x]/hx2 +1i is a ring homomorphism by the mapping a+bi 7−→
(a + bx) + hx2 + 1i.

46
 √
Example
√ A.0.25. ϕ : Q( 2) → Q[x]/hx2 − 2i is a ring homomorphism by the mapping
a + b 2 7−→ (a + bx) + hx2 − 2i.

Polynomial Rings

The following is from Chapter 16, “Polynomial Rings,” in Gallian’s book [6].
Theorem A.0.26. If F is a field then F [x] is a euclidean domain.
Corollary A.0.27. If F is a field, a ∈ F , f (x) ∈ F [x], then f (a) = 0 if and only if
x − a|f (x).
Theorem A.0.28. A polynomial of degree n (over a field) has at most n zeros (counting
multiplicity).
Example A.0.29. In the ring Z12 the polynomial x2 + 8 has four zeros, 2,4,8, and 10.
Theorem A.0.30. Let F be a field. Then F [x] is a principal ideal domain.

Factorization of Polynomials

The following can be found in Chapter 17, “Factorization of Polynomials,” of Gallian’s

book [6].
Theorem A.0.31. Suppose F is a field, f (x) ∈ F [x], and deg(f (x)) ∈ {2, 3}. Then f (x) is
reducible if and only if f (x) has a zero in F .
Theorem A.0.32. If F is a field, and p(x) ∈ F [x], then hp(x)i is maximal if and only if
p(x) is irreducible over F .
Proof. (⇒) Suppose that hp(x)i is maximal in F [x]. We will show that p(x) is irreducible over
F . Let p(x) = g(x)h(x) be a factorization of p(x) over F . Then we have that p(x) ∈ hg(x)i
so hp(x)i ⊆ hg(x)i ⊆ F [x]. Then since hp(x)i is maximal we must have that hp(x)i = hg(x)i
or hp(x)i = F [x].
Suppose that hp(x)i = hg(x)i. Then p(x) and g(x) are associates and they are equal up
to units, so if p(x) = g(x)h(x), we must have that h(x) is a unit, and so p(x) is irreducible.
Suppose that hg(x)i = F [x]. Then we must have that 1 ∈ hg(x)i, thus g −1 (x)g(x) = 1 ∈
hg(x)i and so we must have that g(x) is a unit and so p(x) is irreducible.
(⇐) Suppose that p(x) is irreducible over F , and let I be an ideal of F (x) such that
hp(x)i ⊆ I ⊆ F [x]. We want to show that hp(x)i is maximal by showing that I = f [x] or
I = hp(x)i. Now, since F is a field, we know that F [x] is a principal ideal domain. Thus,
I = hg(x)i for some hg(x)i ∈ F [x]. But, then p(x) ∈ hg(x)i, and so p(x) = g(x)h(x) for

47
some h(x) ∈ F [x]. But then, we know that p(x) is irreducible, so we must have that g(x) is
a unit or h(x) is a unit.
Suppose g(x) is a unit. Then g −1 (x)g(x) = 1 ∈ I, so we must have that I = F [x].
Suppose h(x) is a unit. Then we have that p(x) and g(x) are associates, since they are
equal up to units. Hence, hp(x)i = hg(x)i. And so we have that in either case, hp(x)i is
maximal.

Extension Fields

The following can be found in Chapter 20, “Extension Fields,” of Gallian’s book [6].

Theorem A.0.33. If F is a field and f (x) ∈ F [x] then there exists E such that F ⊆ E and
f (x) has a root in E.

Proof. Suppose that there exists E such that F ⊆ E and f (x) has a root in E. Since F [x]
is a field, it has an irreducible factor, p(x). So, we can let E = F [x]/hp(x)i. Then we have
that the mapping ϕ : F → E by ϕ(a) = a + hp(x)i is injective and per serves multiplication
and addition. Then we want to show that x + hp(x)i is a root of p(x), and thus a root of
f (x). Let p(x) = an xn + an−1 xn−1 + ... + a0 . Then we have

p(x + hp(x)i) = an (x + hp(x)i)n + an−1 (x + hp(x)i)n−1 + ... + a0

= an (xn + hp(x)i) + an−1 (xn−1 + hp(x)i) + ... + a0

= an xn + an−1 xn−1 + ... + a0 + hp(x)i
= p(x) + hp(x)i = 0 + hp(x)i.
Hence, x + hp(x)i is a root of p(x), as desired.

Definition A.0.34. Let E be an extension field of F and let f (x) ∈ F [x] with degree at least
1. We say that f (x) splits in E if there are elements a ∈ F and a1 , a2 , ..., an ∈ E such that

f (x) = a(x − a1 )(x − a2 )...(x − an ).

We call E a splitting field for f (x) over F if

E = F (a1 , a2 , ..., an ).

Algebraic Extensions

The following information is from Chapter 21, “Algebraic Extensions,” of Gallian’s book [6].

48
Definition A.0.35. If E/F (E is a field extension of F ) and ∀u ∈ E we have that f (u) = 0
for some f ∈ F [x] and f 6= 0, then we call this extension an algebraic extension.
√
Example A.0.36. Q( 2)/Q is an algebraic extension.

Example A.0.37. C/R is an algebraic extension.

Example A.0.38. Q(Π)/Q is not algebraic since π is not the root of any polynomial with
rational coefficients.

Theorem A.0.39. Let E/F and u ∈ F be algebraic. Let m be a monic polynomial of

minimum degree such that m(u) = 0. Then

1. m is irreducible over F [x]

2. If f ∈ F [x], then f (u) = 0 iff m|f .

3. m is unique (with respect to u).

Proof. First, we will prove that m is irreducible. Suppose not, so m is reducible and we
have that m(x) = g(x)h(x) where g(x) and h(x) are not units. Then we must have that
m(u) = g(u)h(u) and we know that m(u) = 0, so 0 = g(u)h(u). But, since F is a field, it is
an integral domain and so f (u) = 0 or h(u) = 0, but then m does not have minimal degree
and hence we must have that m is irreducible.
Next, we will show that if f ∈ F [x], then f (u) = 0 iff m|f .
(⇒) Assume f (u) = 0. Then m(u) = f (u). Then we have by the division algorithm that
f (u) = q(u)m(u) + r(u) where deg(r(x)) < deg(m(x)). But then since f (u) = 0, we must
either have that r = 0 or r(u) = 0. But if r(u) = 0, this contradicts that m is the minimal
polynomial, so we must have that r = 0, and thus m|f .
(⇐) Assume that m|f . Then f (x) = m(x)g(x) for some g ∈ F [x]. Thus we have that
f (u) = m(u)g(u) so f (u) = (0)g(u) Thus, f (u) = 0.
Now to show that m is unique suppose it is not. Then there is another irreducible
polynomial, n of minimal degree such that n(u) = 0. But then by (2) we have that m|n and
n|m. Thus m = n and we must have that m is unique.

Definition A.0.40. If u is algebraic over F , then m is called the minimal polynomial of u

over F . We also write dF (u) = deg(m) for the degree of u/F .

Corollary A.0.41. If p ∈ F [x] is monic, irreducible and p(u) = 0, then p is the minimal
polynomial for u.

Theorem A.0.42 (Eisenstein’s Criterion). Let f (x) = an xn + an−1 xn−1 + ... + a1 x + a0 .

Then, if p is prime and

(i) p - an

49
 (ii) p|an−1 , ..., a1 , a0

(iii) p2 - a0
Then f (x) is irreducible over Q.
p √
Example A.0.43. We can check that the minimal polynomial over 1 + 3 is x4 − 2x2 − 2.
p √
We can see that 1 + 3 is a root of x4 − 2x2 − 2, and applying Eisenstein’s Criterion with
p = 2, we have that it is the minimal polynomial.

Theorem A.0.44. Let E/F and u ∈ E be algebraic over F with degF (u) = n. Then,
n−1 
i
P
1. F (u) = ai u |aj ∈ F = {f (u)|f ∈ F [x]}.
i=0

2. {1, u, u2 , ..., un−1 } is a basis for F (u)/F as vector spaces.

3. F (u) ∼
= F [x]/hmi.

50