Leong Yee Hang Geometric Dissection
Leong Yee Hang Geometric Dissection
Leong Yee Hang Geometric Dissection
August 2020
DECLARATION OF ORIGINALITY
I hereby declare that this project report is based on my original work except for
citations and quotations which have been duly acknowledged. I also declare that it
has not been previously and concurrently submitted for any other degree or award at
UTAR or other institutions.
Signature :
ID No. : 1700375
Date : 3/9/2020
ii
APPROVAL FOR SUBMISSION
I certify that this project report entitled “GEOMETRIC DISSECTION” was prepared
by LEONG YEE HANG has met the required standard for submission in partial
fulfilment of the requirements for the award of Bachelor of Science (Honours) Applied
Mathematics With Computing at Universiti Tunku Abdul Rahman.
Approved by,
Signature :
3 September 2020
Date :
iv
The copyright of this report belongs to the author under the terms of the
copyright Act 1987 as qualified by Intellectual Property Policy of University Tunku
Abdul Rahman. Due acknowledgement shall always be made of the use of any material
contained in, or derived from, this report.
vi
ACKNOWLEDGEMENTS
vii
GEOMETRIC DISSECTION
ABSTRACT
viii
TABLE OF C ONTENTS
TITLE i
DECLARATION OF ORIGINALITY ii
ACKNOWLEDGEMENTS vii
ABSTRACT viii
LIST OF FIGURES xi
CHAPTER 1 Introduction 1
1-1 Background & History . . . . . . . . . . . . . . . . . . . 1
1-2 Problem Statement . . . . . . . . . . . . . . . . . . . . . 3
1-3 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . 3
1-4 Notation and Terminology . . . . . . . . . . . . . . . . . 4
1-4-1 Line, Ray and Line Segment . . . . . . . . . . . . 4
1-4-2 Tetrahedron . . . . . . . . . . . . . . . . . . . . . 5
1-5 Project Planning . . . . . . . . . . . . . . . . . . . . . . 6
1-5-1 Project I . . . . . . . . . . . . . . . . . . . . . . . 6
1-5-2 Project II . . . . . . . . . . . . . . . . . . . . . . 6
CHAPTER 4 Equidecomposability 24
4-1 Wallace-Bolyai-Gerwien Theorem . . . . . . . . . . . . . 26
4-2 Hinged Dissection between Any Polygons . . . . . . . . 30
4-3 Hilbert’s Third Problem . . . . . . . . . . . . . . . . . . 32
ix
TABLE OF CONTENTS x
CHAPTER 5 Conclusion 40
5-1 Project Review & Future Study . . . . . . . . . . . . . . 41
L IST OF F IGURES
1.1 Tangram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 T-Puzzle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.3 Stomachion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.4 Hinged Dissection from Triangle to Square . . . . . . . . . . . . . 3
←→
1.5 Line AB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
−→
1.6 Ray AB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
−→
1.7 Ray BA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.8 Line Segment AB . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.9 A Tetrahedron . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
xi
LIST OF FIGURES xii
The history of dissection puzzles can be traced back to the times of Ancient Greek.
Archimedes’ Stomachion is a dissection puzzle similar to Tangram. It has 14 puzzle
pieces which can be arranged into many different shapes like the other dissection
puzzles. However, the main problem associated with Stomachion is the numbers of
different ways of arranging the pieces to form a square. This problem had already
been solved. There are 268 unique arrangements in which no two arrangements are
congruent in terms of rotation and reflection.
Compared to solving the dissection puzzles, creating such interesting dissection
puzzles is generally more challenging. This is because one needs to know how to
dissect a given shape into pieces so that it can be reassembled into another shape. A
mathematical study of this problem is called geometric dissection.
Other than puzzles, geometric dissection could be related to real-life problems.
1
Chapter 1. Introduction 2
Gardner (1977) showed an example: a primitive man had a piece of animal skin, but
it was not in desired shape. He had to find a way to cut it into pieces and sew the
pieces into desired shape. This may be the first dissection problem encountered by a
human. A modern version of this example would be seen when a product is designed
to be in a certain shape but the material is manufactured in some other shape. The
processing plant has to process the material in order to make the product. The problem
is to minimize the processing cost, for instance, by optimizing the number of cuts and
minimizing the waste of material.
A typical example of geometric dissection is Dudeney’s solution (1908) of the
Haberdasher’s puzzle. The puzzle demands to dissect an equilateral triangle into four
pieces such that the four pieces can form a square. The same example also illustrated
hinged dissection. Hinged dissection is a special kind of geometric dissection such that
a number of hinge points can be added to connect the pieces so that transformation into
another shape can be done by rotating the pieces around the hinges. Hinged dissection
was popularised by Dudeney; therefore, it was also known as Dudeney dissection.
An article by Abbott et al. (2012) suggested a good possible application of hinged
dissection, that is building transformable nanobots. Non-hinged dissection is not preferred
for this purpose as it might be difficult to control the transform when the parts are not
physically connected.
Other than dealing with shapes on two-dimensional plane, geometric dissection
can also be done for polyhedra in three-dimensional space. Dissection of polyhedra is
sometimes known as polyhedral dissection. The concept of hinged dissection can be
applied in polyhedral dissection. Instead of having points as hinges, hinged dissection
Chapter 1. Introduction 3
1-3 Objectives
The first objective of this project is to obtain general understanding in geometric
dissection. A clear definition of geometric dissection for both two-dimensional and
three-dimensional space has to be understood. The definition should include conditions
and limitations for dissections and movement of the pieces.
Another objective is to study and research on significant or well-known results
and theorems in geometric dissection. This is mainly done by reading research papers,
journal articles and books. By studying the theorems and proofs, some popular methods
of solving geometric dissection problems can be discovered.
Besides, the project also includes studying of some special geometric dissections.
One of the famous dissections is the hinged dissection. These special dissections
Chapter 1. Introduction 4
usually imposed some conditions upon the regular definition of geometric dissection.
The last objective is to discover new dissections or some new findings. As a
challenge, some open problems related to geometric dissection may be studied. Solving
the problem or part of it will be attempted.
←→
Figure 1.5: Line AB
A ray starting from point A that passes through point B and extends infinitely is
−→ −→ −→
denoted by AB. Note that AB is different from BA. The point that comes first is the
starting endpoint of the ray.
−→
Figure 1.6: Ray AB
−→
Figure 1.7: Ray BA
Chapter 1. Introduction 5
We cannot measure the length of a line or a ray since they extends infinitely but we
can measure the length of a line segment. For AB, we denote it length by |AB|.
1-4-2 Tetrahedron
In this report, we will use the terminologies stated to prevent confusion since some
authors refer (irregular) tetrahedra as "triangular pyramids" and the regular tetrahedron
as "tetrahedron".
Chapter 1. Introduction 6
1-5-1 Project I
Week Plan
1-5-2 Project II
Week Plan
It has been more than a century since geometric dissection is studied extensively by
various famous mathematician. A lot of results regarding geometric dissection have
been published in books, research papers and journal articles. In this chapter, the
resources are studied and some results that have been published are mentioned.
7
Chapter 2. Literature Review 8
2-3 Equidecomposability
Equidecomposability is the term used to describe the possibility of cutting a polygon
into smaller polygons which can be reassembled to form a second polygon. A mathematical
definition of equidecomposability as in Proofs From THE BOOK by Aigner and Ziegler
(2018) is as follows.
A = A1 ∪ A2 ∪ · · · ∪ An
B = B1 ∪ B2 ∪ · · · ∪ Bn
and
Ai is congruent to Bi
for i = 1, 2, ..., n.
A = A1 ∪ A2 ∪ · · · ∪ An
B = B1 ∪ B2 ∪ · · · ∪ Bn
and
Ai is congruent to Bi
for i = 1, 2, ..., n.
The term equidecomposability are used in a number of books and papers including
’A problem of sallee on equidecomposable convex bodies’ by Gardner (1985) and
’Hilbert’s third problem (a story of threes)’ by Krasilnikova (2015). Scissors congruence
is a term which is equivalent to equidecomposability. This term is used in some papers
as well such as Welsh’s expository paper titled ’Scissors congruence’.
Speaking of equidecomposability, the two major questions concerned are:
The answer to the first question is "yes" which is given by the Wallace-Bolyai-
Gerwien Theorem. In terms of equidecomposability, the theorem states
Theorem 2-3-2 (Sydler). Two polyhedra are equildecomposable if and only if they
have equal volume and equal Dehn invariant.
Some books commented that Dehn’s proof was difficult to understand. Some
authors had contributed to rewrite Dehn’s proof. A notable simplification of the proof
Chapter 2. Literature Review 13
a chain, that every piece has exactly two hinges, except for two pieces located at both
ends of the chain which have only one hinge. Furthermore, for every pieces, the edges
which are part of edges of the first polygon should be at the interior of the second
polygon and vice versa. They proved that there exists hinged dissection from
For each of the six results stated, an algorithm to construct the Dudeney dissection
was provided along with the proof. All the results obtained mainly relied on tessellation
of the first polygon. A downside of these algorithms was little or no control over the
output polygon. In some algorithms, there were some steps which a random point
needed to be chosen. This caused the output polygon to have different dimension
in terms of side lengths or angles between the edges even though the algorithm was
Chapter 2. Literature Review 15
started with the same polygon. Also, the algorithms only produce polygons with
similar diameter as the first polygon. In other words, “thin and long” polygon will
be dissected to form another “thin and long” polygon by the algorithm.
CHAPTER 3
D ISSECTION OF S OME P OLYGONS
In this section, we will show and discuss a few dissections of some common 2D shapes
such as rectangles and triangles. Some of the dissection will be analysed in details.
Square is a simple one to begin with. We will first see how to dissect two given
squares to make one larger square. This problem and solution are included in Chapter
2 of Dissections: Plane and Fancy. Let a and b be the side lengths of two squares
with a > b. The construction of this dissection is shown in Figure 3.1. Interestingly,
this dissection can be seen as proof of the Pythagoras theorem. This dissection is due
to Sir George Biddle Airy. Figure 3.2 shows how this dissection can be found using
tessellation.
16
Chapter 3. Dissection of Some Polygons 17
3. Mark midpoint of EC as F .
Now we try to show that this construction is correct. For convenience, let |AB| =
|EB| = x and |BC| = y. Note that the square that we want to construct needs to have
equal area as the rectangle. Therefore, first few steps of the construction are actually
Chapter 3. Dissection of Some Polygons 18
√
aimed to construct a line segment with length xy in order to construct the side length
of the square.
We have
|EC| = x + y
x+y
|F G| = |F E| = |F C| =
2
x+y y−x
|F B| = −x=
2 2
By the Pythagoras theorem, we obtain
s 2 2
x+y y−x √
|BG| = − = xy
2 2
|JA0 | |DM 0 |
=
|JA| |DA|
which simplifies to
√
|JA0 | = xy − x = |DM 0 | − |DC| = |CM 0 |
Construction:
u 1 + √3 2 √
v ! !2
u
√
q
p 3−1 1
|EI| = |EH| = |GH|2 − |GE|2 = t − = 3 = 34
2 2
√
The area of rectangle L0 M K 0 L is equal to the area of the triangle which is 0.5(2)( 3) =
√
3. Rectangle L0 M K 0 L can be proved to be a square by showing that one of its side
p√
has length of 3. In particular, this can be done by showing that |M L0 | = |EI|.
Observe that |M L0 | = |K 0 L| and
2|M L0 | = |M L0 | + |K 0 L|
= |M I 0 | + |I 0 L0 | + |K 0 E| + |EL|
= 2|EI|
Although we know how to construct this dissection starting from the triangle,
constructing the dissection from the other way round, which is to start with a square,
may not be simple. To find such a construction, we start by inspecting the dissection
of square L0 M K 0 L in Figure 3.5. Given such a square, the objective is to construct
the points I 0 , J 0 , E, D and C. The "cuts" that need to made are I 0 J 0 , CD and CE.
Points J 0 and D are midpoints of M K 0 and L0 L respectively. Thus, these two points
1
are relatively simple to construct. When the square has side length 3 4 , it is found that
|I 0 J 0 | = |CD| = |CE| = 1.
Now, the main challenge is to construct the a line segment of length 1 using a line
1
segment of length 3 4 from the starting square. We can apply the technique used in the
1
construction of line segment of length 3 4 from an equilateral triangle of length 2. If
1
we do the same construction on an equilateral triangle of length 3 4 instead, we could
1 1 √
obtain a line segment of length 3 4 /2 · 3 4 = 3/2 instead. We have already known
√ √
that 3 is the height of an equilateral triangle with side length 2, thus 3 is the height
of an equilateral triangle with side length 1. Using this idea, we are able to create the
construction.
Construction:
12. Mark P on BC such that |OP | = |DK|. There are two such possible P , choose
the one which is nearer to B.
The construction steps 2 to 6 are actually the same as those in the construction from
1
equilateral triangle to square. We shall assume |CD| = 3 4 . Based of our arguments
√
earlier on, |DJ| = |F I| would be 3. It can be easily verified that ∠JDK = 30◦ and
∠DKJ = 60◦ . Triangle DJK is a halved equilateral triangle with DJ as height and
DK as an edge. We have successfully constructed a line segment of length 1 which is
DK. By using DK, we can easily construct the "cuts" needed to made on the square.
CHAPTER 4
E QUIDECOMPOSABILITY
This chapter starts with discussion about some properties of equidecomposition. After
that, we will look into the proofs of Wallace-Bolyai-Gerwien Theorem. Finally, we
will study the solution of Hilbert’s Third Problem and its partial proof.
First, let us recall the definition of equidecomposability on page 11. Equidecomposability
is in fact an equivalence relation but this is considered trivial and thus not mentioned
in most books and articles. The reflexive and transitive properties are obviously true,
while the transitivity can be proved easily.
Suppose we have polygons A, B and C such that A is equidecomposable to B and
B is equidecomposable to C. Then, there exists two dissections of polygon B, one
gives the pieces that can be arranged to form polygon A and the other gives the pieces
can be arranged to form polygon C as illustrated in Figure 4.1. By superimposing the
two dissection figures of polygon B, that is, placing one dissection figure of polygon
B on top of another, we observe a new dissection of polygon B as in Figure 4.2. The
pieces produced from this new dissection are able to form polygons A or C when they
are arranged accordingly. Thus, polygons A and C are also equidecomposable.
24
Chapter 4. Equidecomposability 25
by reflection, they can be further dissected so that all pieces are congruent only by
translation and rotation.
Figure 4.3: Gerling’s Dissection from a Triangle to its Reflection (Ciesielska and
Ciesielski, 2018)
The idea of Gerling’s result can be brought into the second dimension. A triangle
can be dissected into 3 pairs of triangular pieces which can be reassembled to form the
reflection of the first triangle as illustrated in Figure 4.3. Such dissection can be easily
constructed by drawing lines where each line passes the incenter and one vertex and of
the triangle. The incenter of a triangle is the center of inscribed circle of the triangle.
The incenter can be constructed as the point where the three angle bisectors of three
vertices of the triangle intersect.
With either statement, we can say that any two polygons of equal area are both
equidecomposable to a third polygon, thus the first two polygons are also equidecomposable.
We shall start by proving the first statement with the aid of the following two lemmas.
Proof. Consider a triangle ABC with AB as its longest edge. The dissection of the
triangle to rectangle can be constructed as follows:
3. The pieces ABGF , F EC and EGC can be arranged into a rectangle as shown
in Figure 4.4
Proof. Consider a non-square rectangle with length l and height h. Without loss of
generality, assume that l is greater than h.
In Section 3-2, we had shown a way to dissect a rectangle to form a square provided
that l 6 4h.
For the case of l > 4h, we show that the rectangle is equidecomposable with a
l0 × h0 rectangle with l0 6 4h0 . We can cut the original rectangle into two rectangles
with length l/2 and height h. These two pieces can be stacked to form a new rectangle
with length l/2 and width 2h. This process can be performed repeatedly to reduce the
length and increase the height of the rectangle as illustrated in Figure 4.5.
Let lk denote the length and hk denote the width of the rectangle obtained after k
rounds of the process. It is easy to see that
k
1
lk = l
2
and
hk = 2k h
and also
lk 1 l
= k
hk 4 h
Chapter 4. Equidecomposability 29
implies
l l 4l
log4 6 n + 1 < log4 + 1 = log4
h h h
which further implies
l 4l
6 4n+1 <
h h
As such, we have
n+1 h
164 <4
l
Taking the reciprocal
1 1 l
< n+1 61
4 4 h
we have
1 l
1< n 64
4 h
implying that
ln
1< 64
hn
and we have
hn < ln 6 4hn
Now the rectangle with length l0 = ln and height h0 = hn satisfies what we want.
Proof. First, dissect the polygon into triangles. By Lemma 4-1-1 the triangles can be
dissected to form rectangles. Next, each small rectangle is dissected to form a square
by Lemma 4-1-2. By using the method mentioned in Section 3-1, we can repeatedly
combine two squares into a larger square until we are left with a single square which
has the same area as the original polygon.
That is the first complete proof for the Wallace-Bolyai-Gerwien Theorem. The
second proof follow the similar arguments as in the paper titled ’Scissors congruence’
by Welsh (2016). The steps are the same as previous proof until the original polygon
is dissected which can form some small rectangles. In the next step, Welsh tried to
prove a lemma stating that any two rectangles of the same area are equidecomposable
Chapter 4. Equidecomposability 30
but there was an incorrect statement in his proof. This mistake was found to be caused
by a wrong construction step of the dissection that comes before that statement. The
construction of the 3-pieces dissection is actually very similar to the dissection from a
rectangle to a square shown in Section 3-2. In fact, we can use this result to prove the
lemma easily.
Lemma 4-1-4. Any two rectangles of the same area are equidecomposable.
Proof. By Lemma 4-1-2, both the rectangles are equidecomposable to a same square
which shares the same area as the two rectangles. Therefore, they are equidecomposable.
Proof. The polygon can be first dissected into triangles and the triangles are further
dissected to form rectangles by Lemma 4-1-1. Next, by Lemma 4-1-4, each of the
rectangle can be dissected to form a rectangle with height 1. Then, all these rectangles
can be stacked to form a long rectangle with height 1 and base equal to sum of the
bases of the rectangular pieces, which obviously has to be numerically equal to a.
Theorem 4-1-6. For a finite set of polygons which all have the same area, there exists
a dissection which the pieces can be arranged to reassemble any polygon from the set.
of hinged dissection does not have a trivial transitive property like equidecomposability.
Even so, (Abbott et al., 2012) had managed to prove the result:
Theorem 4-2-1. For a finite set of polygons which all have the same area, there exists
a hinged dissection in which the pieces connected by hinges can be moved without any
intersection between themselves to reassemble any polygon from the set.
Here we only include the proof outline with brief explanation. The constructive
proof extends from the result in previous section, which is Theorem 4-1-6. The first
step is to add some hinges to the pieces which are from the dissection in Theorem
4-1-6. After adding those hinges, it is not necessary that the pieces with hinges can
reassemble every polygon in the set.
The next step is the most crucial part of the proof. In brief, it is proven that by
further dissecting the current pieces into even smaller pieces which are still connected
by hinges, an original hinge can "change" its position. This idea is illustrated in Figure
4.6.
The method allows the hinges to be "moved" freely and thus a wobbly hinged
dissection that applies for all polygons in the set can be found. To obtain a non-wobbly
hinged dissection, those pieces where intersection occur are further dissected so that
they do not "block the way".
We can observe from Figure 4.6 that there could be a lot of tiny pieces generated.
If the construction is to be built in real world, it could be a challenge for engineers.
We use the dissection in Figure 4.7 to clearly illustrate the concept of segments.
Edge HJ of piece Y consists of two segments as the edge is subdivided by vertices of
other pieces. Meanwhile, edge GH of piece X has the full edge as a single segment.
The same concept can be extended to dissection of polyhedra. The edges of the
polyhedral pieces can be subdivided into segments by vertices or edges of other pieces.
Lemma 4-3-1 (Pearl Lemma). Let P and Q be two polygons or two polyhedra which
are equidecomposable. They are dissected into pieces; P = P1 ∪ ... ∪ Pk , Q =
Q1 ∪ ... ∪ Qk with Pi congruent to Qi . It is possible to put a positive real number
of pearls on the segments such that for each pair of congruent pieces Pi and Qi , the
number of pearls on their corresponding edges are the same.
Figure 4.8 shows a correct assignment of pearls for that dissection but in this lemma
it is not necessary for the number of pearls assigned to be integers.
where xk and yl are the numbers of pearls assigned to the segment sk and s0l respectively.
If we write the linear equation for all edges of every pieces, we have a system of linear
equation
Ax = 0, x>0
where A is an integer matrix with entries 1, −1 or 0 and 0 is the zero vector. Obviously,
a possible solution is the length of each segments and this completes the proof for pearl
lemma.
Lemma 4-3-2 (Cone Lemma). Let A be an integer matrix such that Ax = 0 has a
positive real solution. Then it must has a positive integer solution.
Proof. If the system has a positive real solution, then by multiplying the solution with
a suitable scalar, we can obtain a real solution of at least 1. Therefore, the system
Ax = 0, x>1
where 1 is the vector with all 1’s has a real solution. If we can show that this system has
a rational solution, then we can multiply the solution with the common denominator to
obtain an integer solution which is wanted. Note that we can write the system
Ax = 0, x>1
as
Ax > 0, −Ax > 0, x>1
has a rational solution. We shall prove a more general result, that is,
Ax > b, x>1
Chapter 4. Equidecomposability 35
We can see that there are two independent linear system. By inductive hypotheses,
the first subsystem has a rational solution and in the second subsystem, the smallest
possible xn is chosen similar to what was argued in the base case of n = 1. This
completes the proof.
In fact, the rational solution we obtain in this way is lexicographically smallest in
the solution space. Lexicographical ordering orders the vector based on the first entries
in the vectors. If there is a tie, then those tied vectors are ordered based on their second
entries and so on. This ordering is also called the dictionary ordering since the ordering
of words in a dictionary is similar.
Chapter 4. Equidecomposability 36
Proof. According to the pearl lemma and the cone lemma, we know that it is possible
to assign a positive integer of pearls at the segments of the dissected pieces for both
P and Q such that the corresponding edge for each pair of congruent pieces have the
same number of pearls.
Consider a polyhedral piece P1 dissected from P , every pearl on P1 lies on one of
the edge of P1 . We take the sum of dihedral angle of each pearl and write the sum as
S1 . Suppose P is dissected into k pieces, them we can apply the calculation for the
remaining pieces to obtain S2 , S3 , ..., Sk and let S = S1 + S2 + · · · + Sk .
When the pieces are put together to reassemble P , some pearls from different
pieces may coincide if they share the same segment in P . This concept is illustrated in
Figure 4.9 in 2D. For polyhedra, this situation happens when the pearl lies on a face of
P or at the interior of P . The pearl of the formal case contributes exactly π to S while
a pearl of the latter case contributes exactly 2π to S. The remaining pearls that are not
covered by these two cases must be lying on one of the m edges. Therefore, we can
write
S = a1 α1 + · · · + am αm + cp π
where a1 ,...,am are positive integers (there is at least one pearl on each edge) and cp is a
Chapter 4. Equidecomposability 37
non-negative integer. Since the pieces can also reassemble Q, similar reasoning gives
S = b1 β1 + · · · + bn βn + cq π
where b1 ,...,bn are positive integers and cp is a non-negative integer. We finally arrive
at the Bricard’s condition
a1 α1 + · · · + am αm = b1 β1 + · · · + bn βn + cπ
by taking c = cq − cp .
With Bricard’s condition, now we are able prove that some polyhedra of equal
volume are not equidecomposable. Here we show an example using a square and a
regular tetrahedron. Even if we know that equal volume are necessary for equidecomposability,
we do not really need to care about volume since Bricard’s condition only takes the
dihedral angles into account.
Bricard’s condition tells us that if a square and a regular tetrahedron are equidecomposable,
then
π
12a = 6b(β) + cπ
2
where β is the dihedral angle that is the same for all edges of a regular tetrahedron, a, b
are positive integers and c is a non-negative integer. Observe that left-hand side is a
rational multiple of π. We are done if we can prove that β is not a rational multiple of
π.
We need to find the value of β. Referring to Figure 4.10, let ABCD be a tetrahedron.
E is the center of triangle ABC and ED is to plane ABC. Observe that
which implies
|DF | = |AF | = 3|EF |
Therefore,
|EF | 1
cos(β) = =
|DF | 3
which gives
−1 1
β = cos
3
Chapter 4. Equidecomposability 38
Aigner and Ziegler (2018) has proved in another chapter that for all odd integers n
more than or equal to 3,
1 1
cos−1 √
π n
is irrational. This result can give what we want by using n = 9. We use the argument
from the proof to show that
1 1
cos−1
π 3
is irrational.
We claim that
Nk
cos(kβ) =
3k
where Ak is an integer not divisible by 9 for all non-negative integer k. We prove this
claim by induction. For k = 0 and k = 1, we can easily obtain N0 = N1 = 1. For
k > 2, using the identity
A+B A−B
cos(A) + cos(B) = 2cos cos
2 2
We obtain Nk+1 = 2Nk − 9Nk−1 which is also not divisible by 9 since Ak is not
divisible by 9. The claim is proved.
Assume that
a
β= π
b
for some positive integers a, b. Then
bβ = aπ
gives
cos(bβ) = cos(aπ) = ±1
±3
cos(β) = = ±1
3
This is a contradiction. Thus β is not a rational multiple of π. This completes the proof
that a square and a tetrahedron are not equidecomposable.
CHAPTER 5
C ONCLUSION
The study of geometric dissection starts from investigating how to dissect a polygon
into pieces to other polygons through geometry. In this project, the dissections between
some commonly seen polygons are studied and analysed in details. Some dissections
are trivial. Some examples are the dissection of two squares into one large square
and from a triangle to a rectangle. Meanwhile, some dissection are relatively harder
to discover. The Haberdasher’s problem which asks for a dissection of a equilateral
triangle that form a square is an example.
Hinged dissection or Dudeney dissection is a special kind of dissection which has
an additional requirement. In this dissection, all the pieces are connected by some
hinges and the movement of the pieces are restricted to rotation around the hinges.
If some pieces intersect during the transformation of the pieces from one polygon to
another polygon, this dissection is called a wobbly hinged dissection.
If it is possible to dissect a polygon into finite polygonal pieces and use the pieces to
reassemble another polygon, then these two polygons are said to be equidecomposable
or scissors congruent. This definition applies for polyhedra analogously.
Wallace-Bolyai-Gerwien Theorem states an important result that any two polygons
are equidecomposable if and only if the two polygons have the same area. In other
words, area is the only invariant for dissection of polygons and having equal area is
sufficient and necessary to guarantee equidecomposable of polygons. The constructive
proof of this theorem provided a way to dissect a polygon to reassemble another given
polygon. However, the dissection produces a lot more pieces than required in most
of the cases. In general, finding the minimal dissection between two given polygons
remains as a difficult problem. A recent paper proved a stronger result that hinged
dissection is possible between polygons of equal area. The proof of this finding extends
from the Wallace-Bolyai-Gerwien Theorem.
The generalisation of Wallace-Bolyai-Gerwien Theorem into third dimension, which
would state that equidecomposability of two polyhedra is decided by volume only, is
false. This is generally referred as Hilbert’s Third Problem. Dehn is considered the first
40
Chapter 5. Conclusion 41
person who solved this problem. He proved that there exists a second invariant called
the Dehn invariant for dissection of polyhedra. Sylder later proved that equal volume
and equal Dehn invariant are sufficient to guarantee equidecomposability between two
polygons. An alternative solution of Hilbert’s Third Problem is based on the Bricard’s
condition. Different from Dehn’s solution which is based on abstract algebra, the
solution by Bricard’s condition requires only elementary mathematics.
Abbott, T. G., Abel, Z., Charlton, D., Demaine, E. D., Demaine, M. L. and Kominers,
S. D., 2012. ‘Hinged dissections exist’, Discrete Comput. Geom. 47(1), 150–186.
Aigner, M. and Ziegler, G. M., 2018. Proofs from THE BOOK, Springer.
Brooks, R. L., Smith, C. A., Stone, A. H. and Tutte, W. T., 1940. ‘The dissection of
rectangles into squares’, Duke Math. J. 7(1), 312–340.
Dudeney, H. E., 1908. The Canterbury Puzzles (and Other Curious Problems).
Frederickson, G. N., 2003. Dissections: Plane and Fancy, Cambridge University Press.
Károlyi, G. and Lovász, L., 1991. ‘Decomposition of convex polytopes into simplices’,
preprint 6.
42