#LyX 1.3 created this file. For more info see http://www.lyx.org/
\lyxformat 221
\textclass article
\language english
\inputencoding auto
\fontscheme default
\graphics default
\paperfontsize default
\spacing single
\papersize Default
\paperpackage a4
\use_geometry 0
\use_amsmath 0
\use_natbib 0
\use_numerical_citations 0
\paperorientation portrait
\secnumdepth 3
\tocdepth 3
\paragraph_separation indent
\defskip medskip
\quotes_language english
\quotes_times 2
\papercolumns 1
\papersides 1
\paperpagestyle default
\layout Title
\noun on
Paperfolding
\noun default
\noun on
Sequences
\newline
and quasicrystals
\layout Author
David Longbottom
\layout Abstract
The purpose of this project is to look at the paper-folding sequence, and
how aperiodic sequences can be used in the study of quasicrystals.
I start by describing what the paper-folding sequence is and how to generate
it.
I then show how the sequence is used to draw dragon curves, and discuss
some of the curve's properties.
I talk about an application of one of the properties of the sequence in
the expansion of continued fractions before moving on to describing aperiodic
sequences and their use in quasicrystal theory.
The appendix describes the computer code created for this project.
\layout Standard
\begin_inset LatexCommand \tableofcontents{}
\end_inset
\layout Section
Introduction
\layout Subsection
Outline
\layout Standard
This project will investigate some of the interesting mathematical properties
of paperfolding sequences.
This will include an analysis of how the paperfolding sequence is generated,
its fractal nature, and its relation to continued fractions and other aperiodic
sequences such as the Fibonacci sequence and Rudin-Shapiro sequences.
Then we will discuss how they may be used in the study of one dimensional
quasicrystals.
\layout Subsection
Introduction
\layout Standard
It is common knowledge that a piece of paper cannot be folded in half more
than eight times.
It is less well known that the actual maximum is 12 times
\begin_inset LatexCommand \cite{maxpaper}
\end_inset
.
This project takes as its model the idea of folding a long strip off paper
in half, several times over.
When the paper is unfolded, a series of hills and valleys are revealed.
It is this sequence of hills and valleys that will be studied.
Note, that the limit on how many times a real piece of paper can be folded
is not the point here, and will be ignored.
We assume that we have an
\begin_inset Quotes eld
\end_inset
ideal
\begin_inset Quotes erd
\end_inset
piece of paper that we can fold as many times as we like.
\begin_inset Float figure
placement h
wide false
collapsed true
\layout Standard
\begin_inset Graphics
filename figures/paper_example.fig
\end_inset
\layout Caption
\begin_inset LatexCommand \label{cap:Paper-after-unfolding.}
\end_inset
Example of Paper after unfolding.
\end_inset
\layout Section
Notation
\layout Subsection
Language
\layout Standard
Throughout this paper,
\emph on
folding
\emph default
will be used to describe the act of folding the paper, while
\emph on
creases
\emph default
will be used to describe the series of folds left in the paper when unfolded.
\layout Subsection
Alphabets and Words
\layout Standard
\begin_inset LatexCommand \label{sub:Alphabets-&-Words}
\end_inset
The sequence of
\begin_inset Quotes eld
\end_inset
hills & valleys
\begin_inset Quotes erd
\end_inset
is a binary sequence.
Thus it is natural to represent the sequence by a word made up from an
alphabet of only two letters.
Let
\begin_inset Formula $A$
\end_inset
be an alphabet (i.e.
a finite set), and let
\begin_inset Formula $A^{*}$
\end_inset
be the set of all words made up of letters from
\begin_inset Formula $A$
\end_inset
.
Thus a paperfolding sequence is a subset of
\begin_inset Formula $A^{*}$
\end_inset
.
In this project we shall usually use
\begin_inset Formula $A=\{0,1\}$
\end_inset
, though sometimes
\begin_inset Formula $A=\{-1,+1\}$
\end_inset
will be looked at, depending on which is most appropriate.
So, the above diagram (figure
\begin_inset LatexCommand \ref{cap:Paper-after-unfolding.}
\end_inset
) could be represented by the word
\begin_inset Formula $0010011\in A^{*}$
\end_inset
.
Note that we could also use alphabets like
\begin_inset Formula $\{ a,b\}$
\end_inset
or
\begin_inset Formula $\{\vee,\wedge\}$
\end_inset
or even
\begin_inset Formula $\{\textrm{jack},\textrm{jill}\}$
\end_inset
, it doesn't really matter as they are simply labels.
It is useful to sometimes use the alphabets made of numbers though, especially
for when we come to functional equations.
\layout Standard
The folding instructions (and unfolding instructions) can also be represented
by a similar binary sequence, usually using the same alphabet.
\layout Subsection
Polynomials
\layout Standard
It can also be useful to represent the sequence (of length
\begin_inset Formula $n$
\end_inset
) by a polynomial of the form:
\begin_inset Formula \[
f(x)=\sum_{k=0}^{n-1}a_{k}x^{k}=a_{0}+a_{1}x+a_{2}x^{2}+\cdots+a_{n}x^{n-1}\]
\end_inset
\layout Standard
where
\begin_inset Formula $(a_{k})\epsilon\{0,1\}$
\end_inset
or alternatively
\begin_inset Formula $(a_{k})\epsilon\{-1,+1\}$
\end_inset
.
(See section
\begin_inset LatexCommand \ref{sub:Alphabets-&-Words}
\end_inset
) In this case, for each element in the polynomial, the power of
\begin_inset Formula $x$
\end_inset
represents the position of its coefficient in the word describing the sequence.
So
\begin_inset Formula $0010011\equiv f(x)=x^{2}+x^{5}+x^{6}$
\end_inset
.
Alternatively, if we use the alphabet
\begin_inset Formula $\{-1,+1\}$
\end_inset
, we have the polynomial
\begin_inset Formula $f(x)=-1-x+x^{2}-x^{3}-x^{4}+x^{5}+x^{6}$
\end_inset
.
\newline
We can convert between the two systems.
Say
\begin_inset Formula $f(x)$
\end_inset
uses coefficients
\begin_inset Formula $a_{k}\epsilon\{0,1\}$
\end_inset
, then the equivalent function with coefficients
\begin_inset Formula $\{-1,+1\}$
\end_inset
is
\begin_inset Formula $\sum_{k=0}^{n-1}(-1)^{a_{k}+1}x^{k}$
\end_inset
.
To go the other way (with
\begin_inset Formula $f(x)$
\end_inset
using coefficients
\begin_inset Formula $b_{k}\epsilon\{-1,+1\}$
\end_inset
) we get
\begin_inset Formula $\sum_{k=0}^{n-1}\frac{1}{2}(1+b_{k})x^{k}$
\end_inset
.
\layout Section
Folding Operation
\layout Standard
When the paper is folded in half, an alternating series of creases is placed
between the existing folds, as we see in section
\begin_inset LatexCommand \ref{sub:Folding-Algorithm}
\end_inset
.
Whether this alternating series of folds begins with a 0 or a 1 depends
on which way the paper is folded.
Remember, one can either fold the paper right over left, or left over right.
Also, one can simply turn the paper around or over between each folding.
Thus, the way we fold the paper can be described by a series of folding
instructions
\begin_inset Formula $\in A^{*}$
\end_inset
For the moment, we shall only consider folding the paper the same way each
time, i.e.
the folding instructions are
\begin_inset Formula $0000\ldots$
\end_inset
.
\layout Subsection
Folding Algorithm
\layout Standard
\begin_inset LatexCommand \label{sub:Folding-Algorithm}
\end_inset
It can be seen that the following sequence of creases are observed for each
fold:
\begin_inset Float table
placement H
wide false
collapsed true
\layout Caption
Folding
\layout Standard
\begin_inset Tabular
|
\begin_inset Text
\layout Standard
number of folds
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
|
\begin_inset Text
\layout Standard
2
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
|
\begin_inset Text
\layout Standard
3
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\end_inset
\end_inset
\layout Standard
If one imagines that one side (say the top side) of the paper is black,
and the other side is white, one can then see that folding the paper will
cause it to alternate black/white side up.
That is, a new fold will always cause a same-coloured face to fold onto
itself, so the first half of the side will now be one way up, and the second
half the opposite way up.
So, given this stack of alternating faces up, the folding operation would
cause new creases in all these faces.
The white faces would obtain creases in one direction, and the black faces
would obtain creases in the opposite direction.
Since the white and black faces are alternating, the new creases will alternate
in the same way.
\layout Standard
So a folding instruction of 0 would inserts 0 1 0 1 0 1 ....
between the existing creases, while a folding instruction of 1 would insert
1 0 1 0 1 0 ...
between the existing creases.
\layout Subsubsection
\begin_inset LatexCommand \label{sub:Total-Number-of}
\end_inset
Total Number of Creases in a sequence
\layout Standard
Given
\begin_inset Formula $l_{n}$
\end_inset
creases in the
\begin_inset Formula $n$
\end_inset
-th fold, the
\begin_inset Formula $(n+1)$
\end_inset
th fold will have
\begin_inset Formula $l_{n}+1$
\end_inset
more creases than in the
\begin_inset Formula $n$
\end_inset
-th.
This is because one new crease will appear to the left of each old crease,
and a final one to the right of the last old crease.
So we have the recurrence relation
\begin_inset Formula $l_{n+1}=2l_{n}+1$
\end_inset
.
Since, on no folds we have no creases,
\begin_inset Formula $l_{0}=0$
\end_inset
.
Solving this recurrence relation gives
\begin_inset Formula $l_{n}=2^{n}-1$
\end_inset
\begin_inset Foot
collapsed true
\layout Standard
\begin_inset Formula $l_{n}=2l_{n-1}+1=2(2l_{n-2}+1)+1=4l_{n-2}+2+1=\ldots=2^{n}l_{0}+2^{n-1}+2^{n-2}\ldots+4+2+1=2^{n}l_{0}+2^{n}-1=2^{n}-1$
\end_inset
\end_inset
.
\layout Standard
Note - this means that the number of creases is always odd, and (except
for the zeroth fold) the number of new creases is always even, with an
equal number of ones and zeros added.
\layout Subsubsection
The Algorithm
\layout Standard
\begin_inset LatexCommand \label{sub:algoritm1}
\end_inset
Given a sequence of creases, another fold would introduce creases in between
the existing creases.
Each existing crease would mean the the new creases on either side of it
were opposite in nature.
Our folding instruction defines the first new crease, and the subsequent
new creases will alternate in sign, until we reach the last new crease,
which will be opposite in sign from the first new crease
\begin_inset Foot
collapsed true
\layout Standard
Except for the first fold, which simply introduces a single 0 or 1 (depending
on which way one folds it)
\end_inset
.
\layout Subsection
Folding Operators
\layout Standard
\begin_inset LatexCommand \label{sub:Folding-Operators}
\end_inset
We can now express folding as the following folding operators:
\begin_inset Formula $P_{0},P_{1}:A^{*}\longrightarrow A^{*}$
\end_inset
for
\begin_inset Formula $n>0$
\end_inset
defined as
\begin_inset LatexCommand \cite{fibonaccipdf}
\end_inset
:
\layout Standard
\begin_inset Formula $P_{0}(a_{0}\cdots a_{n})=0a_{0}1a_{1}0\cdots a_{n}1$
\end_inset
\layout Standard
\begin_inset Formula $P_{1}(a_{0}\cdots a_{n})=1a_{0}0a_{1}1\cdots a_{n}0$
\end_inset
\layout Standard
Where
\begin_inset Formula $P_{i}$
\end_inset
is a folding instruction of
\begin_inset Formula $i$
\end_inset
.
(Trivially,
\begin_inset Formula $P_{i}(\emptyset)=i$
\end_inset
) Thus the folding instructions
\begin_inset Formula $101$
\end_inset
can be executed as
\begin_inset Formula $P_{1}\circ P_{0}\circ P_{1}\circ\emptyset\equiv P_{1}(P_{0}(P_{1}(\emptyset)))=P_{1}(P_{0}(1))=P_{1}(011)=1001110$
\end_inset
\layout Subsection
Polynomial Equations
\layout Standard
Given a sequence
\begin_inset Formula $f(x)=a_{0}+a_{1}x+\ldots+a_{n}x^{n}$
\end_inset
, the folding operator will insert new folds between the existing, so we
separate out the existing folds using
\begin_inset Formula $xf(x^{2})=a_{0}x+a_{1}x^{3}+\ldots+a_{n}x^{2n+1}$
\end_inset
.
Now what we do depends on which way we fold, and which alphabet we are
using (see section
\begin_inset LatexCommand \ref{sub:Alphabets-&-Words}
\end_inset
):
\layout Subsubsection
\begin_inset Formula $A=\{0,1\}$
\end_inset
\layout Standard
\begin_inset LatexCommand \label{sub:-A={0,1}}
\end_inset
In the case of a
\begin_inset Formula $0$
\end_inset
-fold we add
\begin_inset Formula $0+x^{2}+0+x^{6}+0+\ldots+0+x^{2n+2}=\sum_{k=0}^{n/2}x^{4k+2}$
\end_inset
to
\begin_inset Formula $xf(x)$
\end_inset
.
In the case of a
\begin_inset Formula $1$
\end_inset
-fold, we add
\begin_inset Formula $1+x^{4}+\ldots+0+x^{2n}+0=\sum_{k=0}^{n/2}x^{4k}$
\end_inset
.
So
\begin_inset Formula $T_{0}(f(x))=xf(x^{2})+\sum_{k=0}^{n/2}x^{4k+2}$
\end_inset
and
\begin_inset Formula $T_{1}(f(x))=xf(x^{2})+\sum_{k=0}^{n/2}x^{4k}$
\end_inset
.
\layout Subsubsection
\begin_inset Formula $A=\{-1,+1\}$
\end_inset
\layout Standard
instead, we would add
\begin_inset Formula $-1+x^{2}-x^{4}+\cdots-x^{2n}+x^{2n+2}=\sum_{k=0}^{n+1}(-1)^{k}x^{2k}$
\end_inset
in the case of a
\begin_inset Formula $0$
\end_inset
-fold.
In the case of a
\begin_inset Formula $1$
\end_inset
-fold we add
\begin_inset Formula $1-x^{2}+x^{4}-\cdots+x^{2n}-x^{2n+2}=\sum_{k=0}^{n+1}-(-1)^{k}x^{2k}$
\end_inset
.
So
\begin_inset Formula $T_{-1}(f(x))=xf(x^{2})+\sum_{k=0}^{n}(-1)^{k}x^{2k}$
\end_inset
and
\begin_inset Formula $T_{+1}(f(x))=xf(x^{2})-\sum_{k=0}^{n}(-1)^{k}x^{2k}$
\end_inset
\layout Subsection
Building up the sequence
\layout Standard
\begin_inset LatexCommand \label{sub:algoritm2}
\end_inset
The paperfolding sequence becomes interesting once we start to look at the
different ways in which it can be built up.
For repeated 0-folds, the sequence looks like this:
\layout Standard
\begin_inset Float table
placement h
wide false
collapsed true
\layout Caption
\begin_inset LatexCommand \label{cap:Folds-&-Seqence}
\end_inset
Folds and Sequences
\layout Standard
\begin_inset Tabular
|
\begin_inset Text
\layout Standard
folds
\end_inset
|
\begin_inset Text
\layout Standard
sequence
\end_inset
|
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $\emptyset$
\end_inset
\end_inset
|
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
|
\begin_inset Text
\layout Standard
2
\end_inset
|
\begin_inset Text
\layout Standard
001
\end_inset
|
|
\begin_inset Text
\layout Standard
3
\end_inset
|
\begin_inset Text
\layout Standard
0010011
\end_inset
|
|
\begin_inset Text
\layout Standard
4
\end_inset
|
\begin_inset Text
\layout Standard
001001100011011
\end_inset
|
|
\begin_inset Text
\layout Standard
5
\end_inset
|
\begin_inset Text
\layout Standard
0010011000110110001001110011011
\end_inset
|
\end_inset
\end_inset
Looking at the pattern indicates another algorithm that can be used to
build up the paperfolding sequence.
It appears that when we fold the paper, the next sequence will start with
the old sequence, add a 0, and then the old sequence reversed and inverted.
So, say
\begin_inset Formula $\overrightarrow{w_{n}}$
\end_inset
is word describing the sequence after the
\begin_inset Formula $n$
\end_inset
-th fold, and
\begin_inset Formula $\overleftarrow{w_{n}}^{\dagger}$
\end_inset
is the mirrored inverse of it (e.g.
\begin_inset Formula $\overleftarrow{001}^{\dagger}=\overrightarrow{011}$
\end_inset
).
\begin_inset Formula \[
\overrightarrow{w_{n+1}}=\overrightarrow{w_{n}}0\overleftarrow{w_{n}}^{\dagger}\]
\end_inset
\layout Standard
To explain this, instead of folding our paper a finite number of times,
we could unfold it a finite number of times instead.
So, after the
\begin_inset Formula $n$
\end_inset
-th unfolding, we have the word
\begin_inset Formula $\overrightarrow{w}$
\end_inset
say.
Then, upon unfolding again, we would have the word
\begin_inset Formula $\overrightarrow{w}0\overleftarrow{w}^{\dagger}$
\end_inset
(assuming, without loss of generality, that this was a
\begin_inset Formula $0$
\end_inset
-fold).
The pattern in table
\begin_inset LatexCommand \ref{cap:Folds-&-Seqence}
\end_inset
is explained as follows: if we fold the same way each time (i.e.
folding instructions 0000...), on the n-th fold, the left hand side will behave
as if it's been folded n-1 times.
\layout Standard
We can generalise this idea, and define an operator
\begin_inset Formula $p_{i}:A^{*}\longrightarrow A^{*}$
\end_inset
,
\begin_inset Formula $p_{i}(\overrightarrow{w})=\overrightarrow{w}i\overleftarrow{w}^{\dagger}$
\end_inset
where
\begin_inset Formula $i\in A^{*}$
\end_inset
is a word.
So a
\begin_inset Formula $0$
\end_inset
-fold is the operation
\begin_inset Formula $p_{0}$
\end_inset
and a
\begin_inset Formula $1$
\end_inset
-fold is the operation
\begin_inset Formula $p_{1}$
\end_inset
.
These will be the same as the operators from section
\begin_inset LatexCommand \ref{sub:Folding-Operators}
\end_inset
, but in reverse order.
\layout Subsection
Functional Equation
\layout Standard
We have observed that the first half of the sequence for
\begin_inset Formula $n+1$
\end_inset
folds is the same as the sequence for
\begin_inset Formula $n$
\end_inset
folds.
This can be expressed using the polynomial notation
\begin_inset Formula $\{0,1\}$
\end_inset
, and
\begin_inset Formula $f_{0}(x)=0$
\end_inset
:
\layout Standard
\begin_inset Formula \[
f_{n+1}(x)=f_{n}(x)+x^{2^{n+1}}\left(\frac{x^{-2^{n+1}+2}-x}{1-x}-f_{n}(x^{-1})\right)\]
\end_inset
\begin_inset Foot
collapsed true
\layout Standard
note
\begin_inset Formula $\frac{x^{-2^{n+1}+2}-x}{1-x}=x^{-(2^{n+1}-2)}+x^{-(2^{n+1}-2)+1}+\cdots+x^{2}+x+1$
\end_inset
where the length of the sequence is
\begin_inset Formula $2^{n+1}-1$
\end_inset
\end_inset
or with
\begin_inset Formula $\{-1,+1\}$
\end_inset
,
\begin_inset Formula $f_{0}(x)=-1$
\end_inset
:
\layout Standard
\begin_inset Formula \[
f_{n+1}(x)=f_{n}(x)-x^{2^{n}}-x^{2^{n+1}}f_{n}(x^{-1})\]
\end_inset
So it is possible, under folding instructions of 000...
(or 111....), to let
\begin_inset Formula $n$
\end_inset
go to infinity.
That is
\begin_inset Formula $\lim_{n\rightarrow\infty}f_{n}(x)=f(x)$
\end_inset
.
If we apply the folding operation to
\begin_inset Formula $f(x)$
\end_inset
we will obtain
\begin_inset Formula $f(x)$
\end_inset
as
\begin_inset Formula $f(x)$
\end_inset
is a stationary point of the folding operation.
\layout Standard
\begin_inset Float table
placement H
wide false
collapsed true
\layout Caption
Observation 2
\layout Standard
\begin_inset Tabular
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $f(x)$
\end_inset
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $\equiv$
\end_inset
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
...
\end_inset
|
|
\begin_inset Text
\layout Standard
-
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $f(x^{2})$
\end_inset
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $\equiv$
\end_inset
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
...
\end_inset
|
|
\begin_inset Text
\layout Standard
=
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $x^{2}+x^{6}+\dots$
\end_inset
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $\equiv$
\end_inset
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
...
\end_inset
|
\end_inset
\end_inset
That is, in functional notation,
\layout Standard
\begin_inset Formula \[
f(x)-f(x^{2})=x^{2}+x^{6}+...=\frac{x^{2}}{1-x^{4}}\]
\end_inset
\layout Standard
As
\begin_inset Formula $(1-x^{4})(x^{2}+x^{6}+x^{10}+\dots)=x^{2}-x^{6}+x^{6}-x^{10}+x^{10}+\dots=x^{2}$
\end_inset
.
This equation is the same as the one in section
\begin_inset LatexCommand \ref{sub:-A={0,1}}
\end_inset
, but with the limit
\begin_inset Formula $n\rightarrow\infty$
\end_inset
.
\layout Subsection
Automata
\layout Standard
The above fact means that, if we 0-fold each time, the sequence of creases
will start the exact same way after each fold.
This indicates that it may be possible to generate the creases sequence
using a automata or a tag machine.
\layout Subsubsection
2-Automaton
\layout Standard
\begin_inset LatexCommand \label{sub:2-Automaton}
\end_inset
The total number of creases in a sequence is given by
\begin_inset Formula $l_{n}=2^{n}-1$
\end_inset
(see section
\begin_inset LatexCommand \ref{sub:Total-Number-of}
\end_inset
).
This indicates that it would be a good idea to index each crease in a sequence
in binary.
Thus the first crease is always labeled by
\begin_inset Formula $1$
\end_inset
and the last crease
\begin_inset Formula $11\ldots11$
\end_inset
.
It is useful to associate with each label the number of 0s from the right
of the binary number before the first 1, i.e.
the highest power of 2 this number is divisible by (I will call this value
the
\emph on
order
\emph default
).
\layout Standard
\begin_inset Float table
placement H
wide false
collapsed true
\layout Caption
1 fold
\layout Standard
\begin_inset Tabular
|
\begin_inset Text
\layout Standard
position
\end_inset
|
\begin_inset Text
\layout Standard
crease
\end_inset
|
\begin_inset Text
\layout Standard
order
\end_inset
|
|
\begin_inset Text
\layout Standard
01
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\end_inset
\end_inset
\begin_inset Float table
placement H
wide false
collapsed true
\layout Caption
2 folds
\layout Standard
\begin_inset Tabular
|
\begin_inset Text
\layout Standard
position
\end_inset
|
\begin_inset Text
\layout Standard
crease
\end_inset
|
\begin_inset Text
\layout Standard
order
\end_inset
|
|
\begin_inset Text
\layout Standard
001
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
|
\begin_inset Text
\layout Standard
010
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
|
\begin_inset Text
\layout Standard
011
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\end_inset
\end_inset
\begin_inset Float table
placement H
wide false
collapsed true
\layout Caption
3 folds
\layout Standard
\begin_inset Tabular
|
\begin_inset Text
\layout Standard
position
\end_inset
|
\begin_inset Text
\layout Standard
crease
\end_inset
|
\begin_inset Text
\layout Standard
order
\end_inset
|
|
\begin_inset Text
\layout Standard
0001
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
|
\begin_inset Text
\layout Standard
0010
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
|
\begin_inset Text
\layout Standard
0011
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
|
\begin_inset Text
\layout Standard
0100
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
2
\end_inset
|
|
\begin_inset Text
\layout Standard
0101
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
|
\begin_inset Text
\layout Standard
0110
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
|
\begin_inset Text
\layout Standard
0111
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\end_inset
\end_inset
\begin_inset Float table
placement H
wide false
collapsed true
\layout Caption
4 folds
\layout Standard
\begin_inset Tabular
|
\begin_inset Text
\layout Standard
position
\end_inset
|
\begin_inset Text
\layout Standard
crease
\end_inset
|
\begin_inset Text
\layout Standard
order
\end_inset
|
|
\begin_inset Text
\layout Standard
0001
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
|
\begin_inset Text
\layout Standard
0010
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
|
\begin_inset Text
\layout Standard
0011
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
|
\begin_inset Text
\layout Standard
0100
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
2
\end_inset
|
|
\begin_inset Text
\layout Standard
0101
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
|
\begin_inset Text
\layout Standard
0110
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
|
\begin_inset Text
\layout Standard
0111
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
|
\begin_inset Text
\layout Standard
1000
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
3
\end_inset
|
|
\begin_inset Text
\layout Standard
1001
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
|
\begin_inset Text
\layout Standard
1010
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
|
\begin_inset Text
\layout Standard
1011
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
|
\begin_inset Text
\layout Standard
1100
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
2
\end_inset
|
|
\begin_inset Text
\layout Standard
1101
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
|
\begin_inset Text
\layout Standard
1110
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
|
\begin_inset Text
\layout Standard
1111
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\end_inset
\end_inset
\layout Standard
\begin_inset LatexCommand \label{order}
\end_inset
The order of each crease corresponds to the number of folds since that crease
was introduced.
In polynomial notation, on folding, crease
\begin_inset Formula $ax^{k}$
\end_inset
is moved to
\begin_inset Formula $xa(x^{2})^{k}=ax^{2k+1}$
\end_inset
.
So its position, which was
\begin_inset Formula $k+1$
\end_inset
, will now be
\begin_inset Formula $2(k+1)$
\end_inset
, i.e.
doubled, which is an extra order in binary.
Note that this is true no matter what folding instructions are used.
In fact the order of the number will tell us what instruction was used
on that fold.
If the order is greater than zero, then we know that the creases on either
side were only introduced this last fold
\layout Standard
Let us observe what happens when we divide each binary index by
\begin_inset Formula $2^{o}$
\end_inset
, where
\begin_inset Formula $o$
\end_inset
is the order of the number):
\layout Standard
\begin_inset Float table
placement H
wide false
collapsed true
\layout Caption
Corrected by order
\layout Standard
\begin_inset Tabular
|
\begin_inset Text
\layout Standard
index -
\begin_inset Formula $2^{o}$
\end_inset
\end_inset
|
\begin_inset Text
\layout Standard
crease
\end_inset
|
|
\begin_inset Text
\layout Standard
00 01
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
|
\begin_inset Text
\layout Standard
00 01
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
|
\begin_inset Text
\layout Standard
00 11
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
|
\begin_inset Text
\layout Standard
00 01
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
|
\begin_inset Text
\layout Standard
01 01
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
|
\begin_inset Text
\layout Standard
00 11
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
|
\begin_inset Text
\layout Standard
01 11
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
|
\begin_inset Text
\layout Standard
00 01
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
|
\begin_inset Text
\layout Standard
10 01
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
|
\begin_inset Text
\layout Standard
01 01
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
|
\begin_inset Text
\layout Standard
10 11
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
|
\begin_inset Text
\layout Standard
00 11
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
|
\begin_inset Text
\layout Standard
11 01
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
|
\begin_inset Text
\layout Standard
01 11
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
|
\begin_inset Text
\layout Standard
11 11
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\end_inset
\end_inset
Observation indicates that, if this new index ends in 01, the crease is
0, and if it ends in 11 the crease is 1.
This is because, on the next fold, pairs of hills & valleys are introduced
on either side of the odd existing creases.
01 tells us that this crease was before one of these odd creases, so is
a valley, and 11 indicates that it came after such a crease, so is a hill.
Therefore our 2-automaton must take the binary index as input (reading
right to left), ignore all the zeros until it reaches a 1, then check the
next binary-digit after that to decide what the crease is like:
\layout Standard
\begin_inset Float figure
placement H
wide false
collapsed true
\layout Standard
\begin_inset Graphics
filename /home/pumaman/math-project/figures/2automaton.fig
scale 80
keepAspectRatio
\end_inset
\layout Caption
2-Automaton for 000...
folding
\end_inset
\layout Standard
Example: to find if the 6th crease is a hill or valley.
6 is 00110 in binary.
Note, there are actually an infinite number of zero digits on the left
hand side, we just don't normally write them.
Our automaton reads the digits from right to left.
It begins at the node marked
\begin_inset Quotes eld
\end_inset
Start
\begin_inset Quotes erd
\end_inset
, and reads the
\begin_inset Quotes eld
\end_inset
0
\begin_inset Quotes erd
\end_inset
instruction, so follows the arrow marked
\begin_inset Quotes eld
\end_inset
0
\begin_inset Quotes erd
\end_inset
, and returns to the
\begin_inset Quotes eld
\end_inset
Start
\begin_inset Quotes erd
\end_inset
node.
The next instruction to be read off is a
\begin_inset Quotes eld
\end_inset
1
\begin_inset Quotes erd
\end_inset
, so we follow the
\begin_inset Quotes eld
\end_inset
1
\begin_inset Quotes erd
\end_inset
arrow to the next node (the one with no label).
The next instruction is also a
\begin_inset Quotes eld
\end_inset
1
\begin_inset Quotes erd
\end_inset
, so we follow that arrow to reach the node marked
\begin_inset Quotes eld
\end_inset
hill
\begin_inset Quotes erd
\end_inset
.
From there all the rest of the instructions (0000...) will keep it at this
node, and the automaton's output is that the 6th crease is a hill.
\layout Subsubsection
What about other folding instructions?
\layout Standard
\begin_inset LatexCommand \label{peanoinstr}
\end_inset
Instead of folding in the same direction each time, we could alternate the
folding instructions, i.e.
0101010101...
So the sequence would evolve as follows:
\begin_inset Float table
placement h
wide false
collapsed true
\layout Caption
Alternating Folding Instructions
\layout Standard
\begin_inset Tabular
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
folding instruction
\end_inset
|
\begin_inset Text
\layout Standard
sequence
\end_inset
|
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
|
\begin_inset Text
\layout Standard
2
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
100
\end_inset
|
|
\begin_inset Text
\layout Standard
3
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0110001
\end_inset
|
|
\begin_inset Text
\layout Standard
4
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
100111001000110
\end_inset
|
\end_inset
\end_inset
\layout Standard
In a similar way to before, we have table
\begin_inset LatexCommand \ref{cap:Alternating-Folding-Instructions}
\end_inset
.
It shows what happens after folding the strip 4 times.
The first column indexes the 15 creases, the second column shows the patter
of hills (1) and valleys (0) when using 0101 as the folding instructions,
and the next column when we use 0000.
\begin_inset Quotes eld
\end_inset
Same?
\begin_inset Quotes erd
\end_inset
checks if these two columns give the same result for that crease (XOR).
Order is the order as described above in section
\begin_inset LatexCommand \ref{order}
\end_inset
, and the final column shows what folding instruction was used when that
crease was introduced.
\begin_inset Float table
placement h
wide false
collapsed true
\layout Caption
Alternating Folding Instructions,
\begin_inset Formula $n=4$
\end_inset
\begin_inset LatexCommand \label{cap:Alternating-Folding-Instructions}
\end_inset
\layout Standard
\begin_inset Tabular
|
\begin_inset Text
\layout Standard
crease
\end_inset
|
\begin_inset Text
\layout Standard
instructions 0101
\end_inset
|
\begin_inset Text
\layout Standard
instructions 0000
\end_inset
|
\begin_inset Text
\layout Standard
same?
\end_inset
|
\begin_inset Text
\layout Standard
order,
\begin_inset Formula $o$
\end_inset
\end_inset
|
\begin_inset Text
\layout Standard
folding instruction (
\begin_inset Formula $n-o$
\end_inset
)
\end_inset
|
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
n
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
|
\begin_inset Text
\layout Standard
2
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
y
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
|
\begin_inset Text
\layout Standard
3
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
n
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
|
\begin_inset Text
\layout Standard
4
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
n
\end_inset
|
\begin_inset Text
\layout Standard
2
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
|
\begin_inset Text
\layout Standard
5
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
n
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
|
\begin_inset Text
\layout Standard
6
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
y
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
|
\begin_inset Text
\layout Standard
7
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
n
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
|
\begin_inset Text
\layout Standard
8
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
y
\end_inset
|
\begin_inset Text
\layout Standard
3
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
|
\begin_inset Text
\layout Standard
9
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
n
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
|
\begin_inset Text
\layout Standard
10
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
y
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
|
\begin_inset Text
\layout Standard
11
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
n
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
|
\begin_inset Text
\layout Standard
12
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
n
\end_inset
|
\begin_inset Text
\layout Standard
2
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
|
\begin_inset Text
\layout Standard
13
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
n
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
|
\begin_inset Text
\layout Standard
14
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
y
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
|
\begin_inset Text
\layout Standard
15
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
n
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\end_inset
\end_inset
\layout Standard
So, by looking at the order (see section
\begin_inset LatexCommand \ref{order}
\end_inset
), one knows what folding instruction was used for that fold.
If it was a 0, then one knows that the crease will be the same as that
for the basic 0000...
instructions.
On the other hand, if is was a 1, then the crease will be opposite.
\layout Standard
A folding instruction of 1 can be thought of as folding the paper in an
opposite way to a 0-fold.
Instead it could be thought of as turning the paper over, then folding
the same way as a 0-fold.
Turning the paper over serves to invert the sequence.
i.e., if the sequence is
\begin_inset Formula $\overrightarrow{w}\in\{-1,1\}^{*}$
\end_inset
then turning it over gives
\begin_inset Formula $-\overrightarrow{w}$
\end_inset
.
Then we can apply the second folding algorithm as before (see section
\begin_inset LatexCommand \ref{sub:algoritm2}
\end_inset
): put a zero onto the end of the sequence, and then put the sequence reversed
and inverted onto the end.
\layout Subsection
Another method
\layout Standard
If folding instructions
\begin_inset Formula $i_{1}i_{2}i_{3}\cdots i_{n}\in A^{*}$
\end_inset
are used to create the paperfolding sequence
\begin_inset Formula $a_{1}a_{2}a_{3}\cdots a_{2^{n}-1}\in A^{*}$
\end_inset
with
\begin_inset Formula $A=\{-1,+1\}$
\end_inset
then we know that certain crease
\begin_inset Formula $a_{k}$
\end_inset
will be the same as the folding instruction
\begin_inset Formula $i_{j}$
\end_inset
.
In fact
\begin_inset Formula $a_{2^{k}}=i_{n-k}$
\end_inset
.
\layout Standard
By the first algorithm (section
\begin_inset LatexCommand \ref{sub:Folding-Operators}
\end_inset
), when we fold, the existing creases will move to positions with double
the index.
So
\begin_inset Formula $a_{2^{k}}\rightarrow a_{2^{k+1}}$
\end_inset
.
After
\begin_inset Formula $n$
\end_inset
folds, a crease
\begin_inset Formula $a_{2^{k}}$
\end_inset
was introduced
\begin_inset Formula $k$
\end_inset
folds ago.
So, after only
\begin_inset Formula $n-k$
\end_inset
folds our crease would have been the first crease
\begin_inset Formula $a_{0}$
\end_inset
.
From section
\begin_inset LatexCommand \ref{sub:Folding-Operators}
\end_inset
, we know this would be same as the folding instruction used at that point,
which is instruction
\begin_inset Formula $i_{n-k}$
\end_inset
.
\layout Standard
So the middle crease
\begin_inset Formula $a_{2^{n}/2}=a_{2^{n-1}}=i_{n-(n-1)}=i_{1}$
\end_inset
is the first fold, and the first crease
\begin_inset Formula $a_{1}=a_{2^{0}}=i_{n-0}=i_{n}$
\end_inset
is the latest fold, as expected.
Now, all the creases introduced on fold
\begin_inset Formula $k$
\end_inset
will depend on instruction
\begin_inset Formula $i_{k}$
\end_inset
.
These creases start with the one indexed by
\begin_inset Formula $2^{k}$
\end_inset
, and every
\begin_inset Formula $2^{k+1}$
\end_inset
th after that.
These crease alternate
\begin_inset Formula $(-1)^{j}$
\end_inset
to the first.
This is exactly what was described above in section
\begin_inset LatexCommand \ref{sub:2-Automaton}
\end_inset
.
Therefore:
\begin_inset Formula \[
a_{2^{k}+j2^{k+1}}=(-1)^{j}i_{n-k}\]
\end_inset
\layout Standard
With
\begin_inset Formula $k=0,\ldots,n-1$
\end_inset
and
\begin_inset Formula $j=0,\ldots,2^{n-k-1}-1$
\end_inset
.
Also, given an arbitrary paperfolding sequence
\begin_inset Formula $a_{0}a_{1}a_{2}a_{3}\cdots$
\end_inset
we can determine the unfolding instructions by looking at the
\begin_inset Formula $2^{k}$
\end_inset
th terms, that is the unfolding instructions
\begin_inset Formula $j_{0}j_{1}j_{2}j_{3}\cdots=a_{1}a_{2}a_{4}a_{8}\cdots$
\end_inset
.
\layout Subsection
Tag Machine
\layout Standard
Consider an infinite tape, a reading head (that reads symbols from the tape),
and a writing head (that writes symbols/words to the tape, according to
rules, depending on what is read from the tape).
The reading head moves along the tape, telling the writing head what to
write down.
The symbols written on the tape can then be converted to the paperfolding
sequence (the symbols on the tape are not necessarily binary).
\layout Standard
Because of the properties described in section
\begin_inset LatexCommand \ref{sub:Total-Number-of}
\end_inset
, it makes sense to start the tape off with 0 (or 1), and have the rules
such that two symbols are written for each one read.
That way, each iteration, when the machine reads the new symbols, the machine
will write enough symbols to provide the next paperfolding sequence.
Investigation shows that an alphabet of 4 for the symbols is appropriate:
\layout Standard
\begin_inset Float table
placement H
wide false
collapsed true
\layout Caption
Tag Machine
\layout Standard
\begin_inset Tabular
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
a
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
01
\end_inset
|
\begin_inset Text
\layout Standard
b
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
c
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
|
\begin_inset Text
\layout Standard
001
\end_inset
|
\begin_inset Text
\layout Standard
0011
\end_inset
|
\begin_inset Text
\layout Standard
b
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
a
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
d
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
c
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
|
\begin_inset Text
\layout Standard
0010011
\end_inset
|
\begin_inset Text
\layout Standard
00011011
\end_inset
|
\begin_inset Text
\layout Standard
b
\end_inset
|
\begin_inset Text
\layout Standard
a
\end_inset
|
\begin_inset Text
\layout Standard
b
\end_inset
|
\begin_inset Text
\layout Standard
c
\end_inset
|
\begin_inset Text
\layout Standard
d
\end_inset
|
\begin_inset Text
\layout Standard
a
\end_inset
|
\begin_inset Text
\layout Standard
d
\end_inset
|
\begin_inset Text
\layout Standard
c
\end_inset
|
\end_inset
\end_inset
\layout Standard
So the following rules control the tag machine:
\layout Standard
\begin_inset Formula $a\mapsto bc$
\end_inset
,
\begin_inset Formula $c\mapsto dc$
\end_inset
,
\begin_inset Formula $d\mapsto da$
\end_inset
,
\begin_inset Formula $b\mapsto ba$
\end_inset
\newline
With the change of symbols:
\layout Standard
\begin_inset Formula $a,b\mapsto0$
\end_inset
,
\begin_inset Formula $c,d\mapsto1$
\end_inset
\newline
to obtain the desired paperfolding sequence.
\layout Subsection
Substitution rules
\layout Standard
\begin_inset LatexCommand \label{sub:Substitution-rules}
\end_inset
This is a bit like the tag machine, but instead of extending the sequence
with the new symbols, the new symbols
\emph on
are
\emph default
the next iteration of the sequence (note, this time iterations do not correspon
d to folds).
The paperfolding sequence is described by the function:
\layout Standard
\begin_inset Formula $a\mapsto ab$
\end_inset
,
\begin_inset Formula $b\mapsto cb$
\end_inset
,
\begin_inset Formula $c\mapsto ad$
\end_inset
,
\begin_inset Formula $d\mapsto cd$
\end_inset
\layout Standard
With the change of symbols:
\layout Standard
\begin_inset Formula $a,b\mapsto0$
\end_inset
(valley),
\begin_inset Formula $c,d\mapsto1$
\end_inset
(hill)
\layout Standard
This is illustrated in table
\begin_inset LatexCommand \ref{cap:Substitution-Rules}
\end_inset
:
\begin_inset Float table
placement h
wide false
collapsed true
\layout Caption
Substitution Rules
\begin_inset LatexCommand \label{cap:Substitution-Rules}
\end_inset
\layout Standard
\begin_inset Tabular
|
\begin_inset Text
\layout Standard
iteration
\end_inset
|
\begin_inset Text
\layout Standard
symbols
\end_inset
|
\begin_inset Text
\layout Standard
transformation
\end_inset
|
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $a$
\end_inset
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $ab$
\end_inset
\end_inset
|
\begin_inset Text
\layout Standard
00
\end_inset
|
|
\begin_inset Text
\layout Standard
2
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $abcd$
\end_inset
\end_inset
|
\begin_inset Text
\layout Standard
0010
\end_inset
|
|
\begin_inset Text
\layout Standard
3
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $abcbadcb$
\end_inset
\end_inset
|
\begin_inset Text
\layout Standard
00100110
\end_inset
|
|
\begin_inset Text
\layout Standard
4
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $abcbadcbabcdadcb$
\end_inset
\end_inset
|
\begin_inset Text
\layout Standard
0010011000110110
\end_inset
|
\end_inset
\end_inset
\layout Standard
As each substitution is of length 2, the substitution rules can be converted
into a 2-automaton, as shown in figure
\begin_inset LatexCommand \ref{cap:Alternative-2-Automaton}
\end_inset
:
\begin_inset Float figure
placement h
wide false
collapsed true
\layout Standard
\begin_inset Graphics
filename /home/pumaman/math-project/figures/2automaton2.fig
\end_inset
\layout Caption
Alternative 2-Automaton
\begin_inset LatexCommand \label{cap:Alternative-2-Automaton}
\end_inset
\end_inset
\layout Standard
For example, the fifth value in the sequence is the
\begin_inset Formula $101$
\end_inset
st value in binary.
Starting from the node marked
\emph on
a
\emph default
, and reading the binary instruction right-to-left, our instructions take
us from
\emph on
a
\emph default
to
\emph on
b
\emph default
(1), from
\emph on
b
\emph default
to
\emph on
c
\emph default
(0) and finally from
\emph on
c
\emph default
to
\emph on
d
\emph default
(1).
Since we end up at
\emph on
d
\emph default
, our transformation tells us that this is a hill.
\layout Section
Dragon Curves
\layout Standard
If we take our sequence of creases, and fold them at 90 degree angles, we
find that certain patterns emerge.
\begin_inset Float figure
placement h
wide false
collapsed true
\layout Standard
\begin_inset Graphics
filename /home/pumaman/math-project/figures/fold1dragon.fig
\end_inset
\layout Caption
Folding paper into a dragon curve
\end_inset
\layout Standard
In order to view the complex patters that emerge, we need to fold the paper
many times over.
By using the automata described in section
\begin_inset LatexCommand \ref{sub:2-Automaton}
\end_inset
, we can write a computer program that will create the paperfolding creases.
We can then write another program that will interpret this sequence of
creases as instructions to turn left or right.
\layout Subsection
Fractals
\layout Standard
When we display the images formed, it is startling to see the fractals that
are produced.
\layout Standard
\begin_inset Float figure
placement H
wide false
collapsed true
\layout Standard
\begin_inset Graphics
filename /home/pumaman/math-project/fractal.png
\end_inset
\layout Caption
Fractal
\begin_inset LatexCommand \label{cap:Fractal}
\end_inset
\end_inset
\layout Subsubsection
1st Method for generating dragon curve
\layout Standard
The first method is to take our dragon curve, and to obtain the next iteration,
`fold it' by inserting new corners between the existing ones.
See figure
\begin_inset LatexCommand \ref{cap:Dragon-Folding-1}
\end_inset
.
\begin_inset Float figure
placement h
wide false
collapsed true
\layout Standard
\begin_inset Graphics
filename figures/dragon_folding.fig
lyxscale 50
scale 50
\end_inset
\layout Caption
Dragon Folding 1
\begin_inset LatexCommand \label{cap:Dragon-Folding-1}
\end_inset
\end_inset
\layout Standard
We can see this more clearly by rotating the next iteration by
\begin_inset Formula $\pi/2$
\end_inset
, scaling by
\begin_inset Formula $\sqrt{2}$
\end_inset
, and overlaying it on top of the current iteration, as in figure
\begin_inset LatexCommand \ref{cap:Dragon-Folding-2}
\end_inset
.
\begin_inset Float figure
placement h
wide false
collapsed true
\layout Standard
\begin_inset Graphics
filename figures/dragon_folding2.fig
lyxscale 50
scale 50
keepAspectRatio
\end_inset
\layout Caption
Dragon Folding 2
\begin_inset LatexCommand \label{cap:Dragon-Folding-2}
\end_inset
\end_inset
\layout Standard
That is, the new curve alternates about the previous curve, with the initial
new corner corresponding to the folding instruction.
This is equivalent to the statement made in section
\begin_inset LatexCommand \ref{sub:algoritm1}
\end_inset
.
\layout Subsubsection
2nd Method for generating dragon curve
\begin_inset LatexCommand \label{sub:dragon-fold-method-2}
\end_inset
\layout Standard
One can also extend the dragon curve using the second method: make a copy
of the curve, place it over the old one, and rotate at the end point by
\begin_inset Formula $\pi/2$
\end_inset
in the direction given by the unfolding instruction, see figure
\begin_inset LatexCommand \ref{cap:Dragon-Folding-3}
\end_inset
.
\begin_inset Float figure
wide false
collapsed true
\layout Standard
\begin_inset Graphics
filename figures/dragon_folding3.fig
lyxscale 50
scale 50
keepAspectRatio
\end_inset
\layout Caption
\begin_inset LatexCommand \label{cap:Dragon-Folding-3}
\end_inset
Dragon Folding 3
\end_inset
\layout Standard
As per the properties of the paperfolding sequence detailed above, these
two methods are equivalent.
\layout Subsection
Simple
\layout Standard
\begin_inset LatexCommand \label{sub:Simple}
\end_inset
All dragon curves are simple, that is they have no multiple edges, and are
not closed.
The proof of this is as follows: Let
\begin_inset Formula $d(i)=(x_{i},y_{i})$
\end_inset
be the coordinates of the
\begin_inset Formula $i$
\end_inset
-th corner (the start of the strip of paper is the
\begin_inset Formula $0$
\end_inset
-th corner).
If the curve were not simple, then for some
\begin_inset Formula $m,n\in\mathbb{Z}$
\end_inset
(without loss of generality, let
\begin_inset Formula $m
|
\begin_inset Text
\layout Standard
year
\end_inset
|
\begin_inset Text
\layout Standard
sequence
\end_inset
|
\begin_inset Text
\layout Standard
number of rabbits
\end_inset
|
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $B$
\end_inset
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $A$
\end_inset
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
|
\begin_inset Text
\layout Standard
2
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $AB$
\end_inset
\end_inset
|
\begin_inset Text
\layout Standard
2
\end_inset
|
|
\begin_inset Text
\layout Standard
3
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $ABA$
\end_inset
\end_inset
|
\begin_inset Text
\layout Standard
3
\end_inset
|
|
\begin_inset Text
\layout Standard
4
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $ABAAB$
\end_inset
\end_inset
|
\begin_inset Text
\layout Standard
5
\end_inset
|
\end_inset
\end_inset
\layout Standard
Note, the number of rabbits satisfies the recurrence relation
\begin_inset Formula $f_{k}=f_{k-1}+f_{k-2}$
\end_inset
as
\begin_inset Formula $f_{k-2}$
\end_inset
will give the number of adult rabbits in year
\begin_inset Formula $k-1$
\end_inset
, so the number this year (
\begin_inset Formula $k$
\end_inset
) will be the number last year plus the number of babies born to adult rabbits.
\layout Standard
So we can let the number of adults and number of babies in the sequence
after
\begin_inset Formula $k$
\end_inset
iterations (years) be represented by the vector
\begin_inset Formula $\left(\begin{array}{c}
\textrm{number of adults}\\
\textrm{number of babies}\end{array}\right)_{k}=\left(\begin{array}{c}
f_{k}\\
f_{k-1}\end{array}\right)$
\end_inset
, and therefore the recurrence relation by:
\layout Standard
\begin_inset Tabular
|
\begin_inset Text
\layout Standard
\begin_inset Formula $\left(\begin{array}{c}
f_{k}\\
f_{k-1}\end{array}\right)$
\end_inset
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $=\left(\begin{array}{cc}
1 & 1\\
1 & 0\end{array}\right)\left(\begin{array}{c}
f_{k-1}\\
f_{k-2}\end{array}\right)$
\end_inset
\end_inset
|
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $=\left(\begin{array}{cc}
1 & 1\\
1 & 0\end{array}\right)^{k-1}\left(\begin{array}{c}
f_{1}\\
f_{0}\end{array}\right)$
\end_inset
\end_inset
|
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\end_inset
\layout Standard
The eigenvalues of the matrix
\begin_inset Formula $\left(\begin{array}{cc}
1 & 1\\
1 & 0\end{array}\right)$
\end_inset
are
\begin_inset Formula $\frac{1\pm\sqrt{5}}{2}$
\end_inset
.
Note that
\begin_inset Formula $\phi=\frac{1+\sqrt{5}}{2}$
\end_inset
is the golden ratio.
\layout Standard
Assume
\begin_inset Formula $f_{k}/f_{k-1}$
\end_inset
tends to a limit, i.e.
\begin_inset Formula $\left(\begin{array}{cc}
1 & 1\\
1 & 0\end{array}\right)\left(\begin{array}{c}
f_{k-1}\\
f_{k-2}\end{array}\right)=\left(\begin{array}{c}
f_{k}\\
f_{k-1}\end{array}\right)=\lambda\left(\begin{array}{c}
f_{k-1}\\
f_{k-2}\end{array}\right)$
\end_inset
.
Obviously
\begin_inset Formula $\phi$
\end_inset
is an eigenvalue of the matrix, and as the Fibonacci sequence is positive,
\begin_inset Formula $\lim_{k\rightarrow\infty}f_{k}/f_{k-1}=\frac{1+\sqrt{5}}{2}=\phi$
\end_inset
, i.e.
the golden ratio.
\layout Standard
Also, note that the Fibonacci sequence is a sturmian sequence with
\begin_inset Formula $\alpha=\phi-1=\frac{\sqrt{5}-1}{2}$
\end_inset
and
\begin_inset Formula $\beta=0$
\end_inset
.
\layout Subsection
Rudin-Shapiro
\begin_inset LatexCommand \label{sub:Rubin-shapiro}
\end_inset
\layout Standard
The Rudin-Shapiro sequence gives the parity of the number of
\begin_inset Formula $11$
\end_inset
blocks in a binary number.
For example (if we want the sequence given using
\begin_inset Formula $\{0,1\}$
\end_inset
), the
\begin_inset Formula $27$
\end_inset
th term, which is
\begin_inset Formula $11011$
\end_inset
in binary, has two
\begin_inset Formula $11$
\end_inset
s in it, so has a value of
\begin_inset Formula $0$
\end_inset
in the sequence.
On the other hand
\begin_inset Formula $26$
\end_inset
is
\begin_inset Formula $11010$
\end_inset
in binary, has one
\begin_inset Formula $11$
\end_inset
, so has a value of
\begin_inset Formula $1$
\end_inset
.
See table
\begin_inset LatexCommand \ref{cap:Rudin-Shapiro}
\end_inset
.
\begin_inset Float table
placement h
wide false
collapsed true
\layout Caption
Rudin-Shapiro
\begin_inset LatexCommand \label{cap:Rudin-Shapiro}
\end_inset
\layout Standard
\begin_inset Tabular
|
\begin_inset Text
\layout Standard
term in sequence
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
2
\end_inset
|
\begin_inset Text
\layout Standard
3
\end_inset
|
\begin_inset Text
\layout Standard
4
\end_inset
|
\begin_inset Text
\layout Standard
5
\end_inset
|
\begin_inset Text
\layout Standard
6
\end_inset
|
\begin_inset Text
\layout Standard
7
\end_inset
|
\begin_inset Text
\layout Standard
8
\end_inset
|
\begin_inset Text
\layout Standard
9
\end_inset
|
\begin_inset Text
\layout Standard
10
\end_inset
|
|
\begin_inset Text
\layout Standard
in binary
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
10
\end_inset
|
\begin_inset Text
\layout Standard
11
\end_inset
|
\begin_inset Text
\layout Standard
100
\end_inset
|
\begin_inset Text
\layout Standard
101
\end_inset
|
\begin_inset Text
\layout Standard
110
\end_inset
|
\begin_inset Text
\layout Standard
111
\end_inset
|
\begin_inset Text
\layout Standard
1000
\end_inset
|
\begin_inset Text
\layout Standard
1001
\end_inset
|
\begin_inset Text
\layout Standard
1010
\end_inset
|
|
\begin_inset Text
\layout Standard
number of
\begin_inset Formula $11$
\end_inset
s
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
2
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
|
\begin_inset Text
\layout Standard
modulo 2
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
|
\begin_inset Text
\layout Standard
sequence in
\begin_inset Formula $\{-1,+1\}$
\end_inset
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
-1
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
-1
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\end_inset
\end_inset
\layout Standard
The last row in this table is the Rudin-Shapiro sequence.
It can also be built up by alternating self addition, see table
\begin_inset LatexCommand \ref{cap:Rudin-Shapiro-2}
\end_inset
for an illustration (this time using
\begin_inset Formula $\{1,-1\}$
\end_inset
as the alphabet for the sequence):
\begin_inset Float table
placement h
wide false
collapsed true
\layout Caption
Rudin-Shapiro 2
\begin_inset LatexCommand \label{cap:Rudin-Shapiro-2}
\end_inset
\layout Standard
\begin_inset Tabular
|
\begin_inset Text
\layout Standard
Rudin-Shapiro sequence
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
-1
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
-1
\end_inset
|
|
\begin_inset Text
\layout Standard
alternating self addition
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
-(1)
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
-(-1)
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
-(1)
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
|
\begin_inset Text
\layout Standard
the Rudin-Shapiro sequence again!
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
-1
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
-1
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
-1
\end_inset
|
\begin_inset Text
\layout Standard
-1
\end_inset
|
\end_inset
\end_inset
\layout Standard
This is summarised in the following recurrence relation, where
\begin_inset Formula $r_{k}\in\{-1,1\}$
\end_inset
is the
\begin_inset Formula $k$
\end_inset
th term in the sequence,
\begin_inset Formula $k\geq0$
\end_inset
:
\layout Standard
\begin_inset Formula $r_{0}=1$
\end_inset
\layout Standard
\begin_inset Formula $r_{2k}=r_{k}$
\end_inset
(first line of table
\begin_inset LatexCommand \ref{cap:Rudin-Shapiro-2}
\end_inset
) Note, this also means that
\begin_inset Formula $r_{2^{n}}=r_{0}$
\end_inset
for
\begin_inset Formula $n\geq0$
\end_inset
.
\layout Standard
\begin_inset Formula $r_{2k+1}=(-1)^{k}r_{k}$
\end_inset
(second line of table
\begin_inset LatexCommand \ref{cap:Rudin-Shapiro-2}
\end_inset
)
\layout Standard
This is expressed in the functional equation (for the infinite sequence),
where
\begin_inset Formula $R(x)=\sum_{k=0}^{\infty}r_{k}x^{k}$
\end_inset
,
\layout Standard
\begin_inset Formula \[
R(x)=R(x^{2})+xR(-x^{2})\]
\end_inset
\layout Standard
Or, for the
\begin_inset Formula $n$
\end_inset
th iteration, where
\begin_inset Formula $R_{n}(x)=\sum_{k=0}^{2^{n}-1}r_{k}x^{k}$
\end_inset
, we have the recurrence relation
\layout Standard
\begin_inset Formula \[
R_{n+1}(x)=R_{n}(x^{2})+xR_{n}(-x^{2})\]
\end_inset
\layout Subsubsection
Automation
\layout Standard
This sequence is described by the substitution rules in a similar way to
section
\begin_inset LatexCommand \ref{sub:Substitution-rules}
\end_inset
.
The rules are
\begin_inset LatexCommand \cite{quantum}
\end_inset
:
\layout Standard
\begin_inset Formula $a\mapsto ab$
\end_inset
,
\begin_inset Formula $b\mapsto ac$
\end_inset
,
\begin_inset Formula $c\mapsto db$
\end_inset
,
\begin_inset Formula $d\mapsto dc$
\end_inset
\layout Standard
With the change of symbols:
\layout Standard
\begin_inset Formula $a,b\mapsto0$
\end_inset
(or
\begin_inset Formula $1$
\end_inset
),
\begin_inset Formula $c,d\mapsto1$
\end_inset
(or
\begin_inset Formula $-1$
\end_inset
)
\layout Standard
This has the equivalent automaton shown in figure :
\begin_inset Float figure
placement h
wide false
collapsed true
\layout Standard
\begin_inset Graphics
filename /home/pumaman/math-project/figures/2automaton3.fig
\end_inset
\layout Caption
Rudin-Shapiro 2-Automaton
\end_inset
\layout Subsubsection
Properties
\layout Standard
Consider the exponential sum and the integral
\begin_inset Formula \[
\sum_{k=0}^{n-1}\pm e^{ik\theta}\]
\end_inset
\begin_inset Formula \[
\int_{0}^{2\pi}\left|\sum_{k=0}^{n-1}\pm e^{ik\theta}\right|^{2}d\theta=n\]
\end_inset
\layout Standard
regardless of the choice of sign
\begin_inset Formula $\pm$
\end_inset
.
It then follows that
\begin_inset Formula \[
\sup_{0\leq\theta\leq2\pi}\left|\sum_{k=0}^{n-1}\pm e^{ik\theta}\right|\geq\sqrt{n}\]
\end_inset
\layout Standard
One of the properties of the Rudin-Shapiro sequence
\begin_inset Formula $\{ r_{k}\}$
\end_inset
is that for all
\begin_inset Formula $n$
\end_inset
(using the alphabet
\begin_inset Formula $r_{k}\in\{-1,1\}$
\end_inset
)
\begin_inset LatexCommand \cite{folds}
\end_inset
\begin_inset Formula \[
\sup_{0\leq\theta\leq2\pi}\left|\sum_{k=0}^{n-1}r_{k}e^{ik\theta}\right|\leq(2+\sqrt{2})\sqrt{n}\]
\end_inset
\layout Section
More on Dragon Curves
\layout Subsection
Space filling
\layout Standard
\begin_inset LatexCommand \label{sub:Space-filling}
\end_inset
In section
\begin_inset LatexCommand \ref{Observation-space-filling}
\end_inset
we observed that dragon curves somehow fill out an area of the plane.
We will now formalise this concept.
A normal curve should be one-dimensional, but our dragon curves will, with
a suitable definition of dimension, turn out to be two-dimensional.
\layout Standard
Given a curve
\begin_inset Formula $\Gamma$
\end_inset
, let the set
\begin_inset Formula $\Gamma(\epsilon)=\left\{ y\,:\, distance(y,x)<\epsilon;\forall x\in\Gamma\right\} $
\end_inset
.
This is giving our curve
\begin_inset Formula $\Gamma$
\end_inset
some
\emph on
thickness
\emph default
\begin_inset Formula $\epsilon$
\end_inset
- see figure
\begin_inset LatexCommand \ref{cap:Epsilonised-peano-curve}
\end_inset
.
\begin_inset Float figure
placement h
wide false
collapsed true
\layout Standard
\begin_inset Graphics
filename /home/pumaman/math-project/figures/peanoepsilon.fig
\end_inset
\layout Caption
Epsilonised Peano curve
\begin_inset LatexCommand \label{cap:Epsilonised-peano-curve}
\end_inset
\end_inset
\layout Standard
We will actually consider the limit where
\begin_inset Formula $\Gamma$
\end_inset
is infinitely long (i.e.
is an infinite paperfolding curve).
Now also consider the area common to
\begin_inset Formula $\Gamma(\epsilon)$
\end_inset
and a disk (centered at the origin) of radius
\begin_inset Formula $R$
\end_inset
, and call it
\begin_inset Formula $\Gamma(\epsilon;R)$
\end_inset
.
Then we can define our dimension as:
\begin_inset Formula \[
\textrm{dim}\,\Gamma(\epsilon,R)=\frac{\log\Gamma(\epsilon;R)}{\log R}\]
\end_inset
\layout Standard
where
\begin_inset Formula $\log$
\end_inset
is the natural logarithm, and we take the limits
\begin_inset Formula $\lim_{R\rightarrow\infty}$
\end_inset
and
\begin_inset Formula $\lim\epsilon\rightarrow0$
\end_inset
.
This equation can be rearranged as
\begin_inset Formula $\Gamma(\epsilon;R)=R^{\textrm{dim}\,\Gamma}$
\end_inset
.
This would indicate that
\begin_inset Formula $1\leq\textrm{dim}\,\Gamma\leq2$
\end_inset
as the common area would be between
\begin_inset Formula $R^{1}$
\end_inset
and
\begin_inset Formula $R^{2}$
\end_inset
.
For a simple line,
\begin_inset Formula $\Gamma(\epsilon;R)\approx R$
\end_inset
, so the dimension would be 1.
We will now show that the dimension for a dragon curve is 2.
\layout Standard
First we must consider the dragon curve lying in the complex plane, with
the
\begin_inset Formula $k$
\end_inset
th corner having coordinates
\begin_inset Formula $z_{k}=x_{k}+iy_{k}$
\end_inset
(let
\begin_inset Formula $z_{0}=0$
\end_inset
and
\begin_inset Formula $z_{1}=1$
\end_inset
).
We will now split the turning instructions into a pair of sequences describing
the horizontal and vertical moves .
This is done by keeping track of the cumulative difference in instructions
to turn left and to turn right.
If the paperfolding sequence uses the alphabet
\begin_inset Formula $\{-1,+1\}$
\end_inset
then the count,
\begin_inset Formula $c_{k}=\sum_{j=0}^{k}crease_{j}$
\end_inset
.
The even counts
\begin_inset Formula $c_{2k}$
\end_inset
define the horizontal movements and the odd counts
\begin_inset Formula $c_{2k+1}$
\end_inset
describe the vertical movements.
The count is useful as it can be decoded as follows:
\layout Standard
\begin_inset Tabular
|
\begin_inset Text
\layout Standard
\begin_inset Formula $c_{2k}\equiv0\textrm{mod}4$
\end_inset
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $\rightarrow$
\end_inset
instruction to move right
\end_inset
|
|
\begin_inset Text
\layout Standard
\begin_inset Formula $c_{2k}\equiv2\textrm{mod}4$
\end_inset
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $\leftarrow$
\end_inset
instruction to move left
\end_inset
|
|
\begin_inset Text
\layout Standard
\begin_inset Formula $c_{2k+1}\equiv1\textrm{mod}4$
\end_inset
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $\uparrow$
\end_inset
instruction to move up
\end_inset
|
|
\begin_inset Text
\layout Standard
\begin_inset Formula $c_{2k+1}\equiv3\textrm{mod}4$
\end_inset
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $\downarrow$
\end_inset
instruction to move down
\end_inset
|
\end_inset
\layout Standard
Table
\begin_inset LatexCommand \ref{cap:Count}
\end_inset
shows how the count is calculated, and how it is decoded into the directions
of each segment of the dragon curve.
Note, the first segment is set to be
\begin_inset Formula $\rightarrow$
\end_inset
.
The table also shows what happens when we represent the the horizontal
and vertical movements by the numbers
\begin_inset Formula $0\equiv\rightarrow\equiv\downarrow$
\end_inset
and
\begin_inset Formula $1\equiv\leftarrow\equiv\uparrow$
\end_inset
respectively.
\begin_inset Float table
placement h
wide false
collapsed true
\layout Caption
Count
\begin_inset LatexCommand \label{cap:Count}
\end_inset
\layout Standard
\begin_inset Tabular
|
\begin_inset Text
\layout Standard
sequence
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
-1
\end_inset
|
\begin_inset Text
\layout Standard
+1
\end_inset
|
\begin_inset Text
\layout Standard
+1
\end_inset
|
\begin_inset Text
\layout Standard
-1
\end_inset
|
\begin_inset Text
\layout Standard
-1
\end_inset
|
\begin_inset Text
\layout Standard
-1
\end_inset
|
\begin_inset Text
\layout Standard
+1
\end_inset
|
\begin_inset Text
\layout Standard
+1
\end_inset
|
\begin_inset Text
\layout Standard
-1
\end_inset
|
\begin_inset Text
\layout Standard
+1
\end_inset
|
\begin_inset Text
\layout Standard
+1
\end_inset
|
|
\begin_inset Text
\layout Standard
count
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
-1
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
-1
\end_inset
|
\begin_inset Text
\layout Standard
-2
\end_inset
|
\begin_inset Text
\layout Standard
-1
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
-1
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $\rightarrow$
\end_inset
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $\downarrow$
\end_inset
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $\rightarrow$
\end_inset
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $\uparrow$
\end_inset
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $\rightarrow$
\end_inset
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $\downarrow$
\end_inset
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $\leftarrow$
\end_inset
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $\downarrow$
\end_inset
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $\rightarrow$
\end_inset
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $\downarrow$
\end_inset
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $\rightarrow$
\end_inset
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $\uparrow$
\end_inset
\end_inset
|
|
\begin_inset Text
\layout Standard
horizontal
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
|
\begin_inset Text
\layout Standard
vertical
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
|
\begin_inset Text
\layout Standard
combined
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
0
\end_inset
|
\begin_inset Text
\layout Standard
1
\end_inset
|
\end_inset
\end_inset
\layout Standard
It is interesting to note that the combined sequence is the Rudin-Shapiro
sequence as described in section
\begin_inset LatexCommand \ref{sub:Rubin-shapiro}
\end_inset
.
Even more interestingly, if we let
\begin_inset Formula $\rightarrow=A$
\end_inset
,
\begin_inset Formula $\downarrow=B$
\end_inset
,
\begin_inset Formula $\uparrow=C$
\end_inset
and
\begin_inset Formula $\leftarrow=D$
\end_inset
then we have the Rudin-Shapiro sequence as generated by its substitution
rules (i.e.
before the change of symbols).
\layout Standard
We can now define the direction function
\begin_inset Formula $D(x)$
\end_inset
, the functional equation with real coefficients describing the horizontal
movements, and imaginary ones describing the vertical movements.
If the
\begin_inset Formula $k$
\end_inset
th horizontal move and
\begin_inset Formula $k$
\end_inset
th vertical moves are described by
\begin_inset Formula $h_{k},v_{k}\in\{0,1\}$
\end_inset
respectively, then the count can be decoded by
\begin_inset Formula $i^{c_{2k}}=(-1)^{h_{k}}$
\end_inset
and
\begin_inset Formula $i^{c_{2k+1}}=-i(-1)^{v_{k}}$
\end_inset
.
The direction function is then:
\layout Standard
\begin_inset Formula \[
D(x)=\sum_{k\geq0}i^{c_{k}}x^{k}=\sum_{k\geq0}(-1)^{h_{k}}x^{2k}-i\sum_{k\geq0}(-1)^{v_{k}}x^{2k+1}\]
\end_inset
\layout Standard
So, from the example in table
\begin_inset LatexCommand \ref{cap:Count}
\end_inset
,
\begin_inset Formula $D(x)=1-ix+x^{2}+ix^{3}+x^{4}-ix^{5}-x^{6}-\ldots$
\end_inset
.
Now, if given a paper folding sequence defined by the folding instructions
\begin_inset Formula $(\epsilon_{1},\epsilon_{2},\ldots,\epsilon_{n})$
\end_inset
, where
\begin_inset Formula $\epsilon_{j}\in\{-1,+1\}$
\end_inset
and
\begin_inset Formula $\epsilon_{n}$
\end_inset
was the first folding instruction (or last unfolding instruction), and produces
a sequence of length
\begin_inset Formula $2^{n}-1$
\end_inset
.
This sequence is made up of the previous sequence repeated and folded about
\begin_inset Formula $\epsilon_{1}$
\end_inset
(see section
\begin_inset LatexCommand \ref{sub:dragon-fold-method-2}
\end_inset
).
The reverse of this is a matter of cutting the paper in half (at fold
\begin_inset Formula $\epsilon_{1}$
\end_inset
), thus loosing the first folding instruction.
the second half of the paper is the mirror image of the first, so its central
fold is the reverse of the one in the first half.
So the first half of the sequence given by instructions
\begin_inset Formula $(\epsilon_{1},\epsilon_{2},\ldots,\epsilon_{n})$
\end_inset
is given by
\begin_inset Formula $(\epsilon_{2},\ldots,\epsilon_{n})$
\end_inset
, and the second half by
\begin_inset Formula $(-\epsilon_{2},\ldots,\epsilon_{n})$
\end_inset
.
Each of these halves is a sequence of length
\begin_inset Formula $2^{n-1}-1$
\end_inset
.
So we can write the following recurrence relation for the
\begin_inset Formula $D(x)$
\end_inset
functional equation.
\layout Standard
\begin_inset Formula \[
D(\epsilon_{1},\epsilon_{2},\dots,\epsilon_{n};x)=D(\epsilon_{2},\dots,\epsilon_{n};x)+i^{\epsilon_{1}+\epsilon_{2}}x^{2^{n-1}}D(-\epsilon_{2},\dots,\epsilon_{n};x)\]
\end_inset
\layout Standard
With
\begin_inset Formula $D(\epsilon_{n};x)=1+i^{\epsilon_{n}}x$
\end_inset
(and
\begin_inset Formula $D(\emptyset;x)=1$
\end_inset
).
Note the orientation of the second half is determined by a rotation
\begin_inset Formula $i^{c_{2^{n-1}}}=i^{\epsilon_{1}+\epsilon_{2}}$
\end_inset
.
So
\begin_inset Formula $\epsilon_{1}+\epsilon_{2}=c_{2^{n-1}}$
\end_inset
.
It is interesting to note that we can draw dragon curves with angles other
than
\begin_inset Formula $\pi/2\rightarrow e^{i\pi/2}=i$
\end_inset
by replacing
\begin_inset Formula $i$
\end_inset
in the equation with
\begin_inset Formula $e^{i\theta}$
\end_inset
, where
\begin_inset Formula $\theta$
\end_inset
is the angle.
\layout Standard
The functional equation can represented by the matrix identity:
\layout Standard
\begin_inset Formula \[
\left(\begin{array}{c}
D(\epsilon_{1},\epsilon_{2},\dots,\epsilon_{n};z)\\
D(-\epsilon_{1},\epsilon_{2},\dots,\epsilon_{n};z)\end{array}\right)=\left(\begin{array}{cc}
1 & i^{\epsilon_{1}+\epsilon_{2}}z^{2^{n-1}}\\
1 & -i^{\epsilon_{1}+\epsilon_{2}}z^{2^{n-1}}\end{array}\right)\times\left(\begin{array}{c}
D(\epsilon_{2},\dots,\epsilon_{n};z)\\
D(-\epsilon_{2},\dots,\epsilon_{n};z)\end{array}\right)\]
\end_inset
\layout Standard
By repeatedly halving, we get the identity:
\layout Standard
\begin_inset Formula \[
\left(\begin{array}{c}
D(\epsilon_{1},\epsilon_{2},\dots,\epsilon_{n};z)\\
D(-\epsilon_{1},\epsilon_{2},\dots,\epsilon_{n};z)\end{array}\right)=\prod_{k=1}^{n-1}\left(\begin{array}{cc}
1 & i^{\epsilon_{n-k}+\epsilon_{n-k-1}}z^{2^{k}}\\
1 & -i^{\epsilon_{n-k}+\epsilon_{n-k-1}}z^{2^{k}}\end{array}\right)\times\left(\begin{array}{c}
1+i^{\epsilon_{n}}\\
1-i^{\epsilon_{n}}\end{array}\right)\]
\end_inset
\layout Standard
By setting
\begin_inset Formula $\left|z\right|=1$
\end_inset
, so that
\begin_inset Formula $D(z)$
\end_inset
gives the complex position of the end of the dragon curve, the norm of
the matrix operators on the right are
\begin_inset Formula $\sqrt{2}$
\end_inset
, so the norm
\begin_inset Formula $\left|D(\epsilon_{1},\epsilon_{2},\dots,\epsilon_{n};z)\right|\leq\sqrt{2}^{n}=2^{n/2}$
\end_inset
.
So, the end of the dragon curve is less than or equal to
\begin_inset Formula $\sqrt{2}^{n}$
\end_inset
from the origin.
Now, if we go back to our definition of dimension
\begin_inset Formula $\textrm{dim}\,\Gamma(\epsilon,R)=\ln\Gamma(\epsilon;R)/\ln R$
\end_inset
, we have
\begin_inset Formula $R=2^{n/2}$
\end_inset
, and
\begin_inset Formula $\Gamma(\epsilon;R)=2^{n}$
\end_inset
as there are
\begin_inset Formula $2^{n}$
\end_inset
line segments of length
\begin_inset Formula $1$
\end_inset
(since there are
\begin_inset Formula $2^{n}-1$
\end_inset
creases) in the circle of radius
\begin_inset Formula $R$
\end_inset
.
So
\begin_inset Formula $\textrm{dim}\,\Gamma(\epsilon,R)=\ln2^{n}/\ln2^{n/2}=2\ln2^{n}/\ln2^{n}=2$
\end_inset
, the dimension of a dragon curve is 2, as desired.
\layout Subsection
Dimension
\layout Standard
Mandelbrot defined a fractal to be a set with Hausdorff dimension strictly
exceeding the topological dimension.
This definition is the generally accepted one, though it does fail in some
cases with sets that we would like to be included as a fractal.
\layout Standard
Normally dimension would just mean: zero dimensional points; one dimensional
lines; two dimensional shapes; three dimensional object etc.
It is defined iteratively, that is a three dimensional space is contained
in a two dimensional surface; a two dimensional shape is contained in by
a one dimensional boundary etc.
Other definitions of dimension, however, can have non-integer values.
The motivation for this is seen, for example, if we take the boundary of
our dragon fractal.
It is not one-dimensional as if we measure the length of the boundary it
is infinite, but it is nether two-dimensional as if we measure the area
of the boundary, it is zero! Compare this, say, with a square.
It has a one dimensional boundary as we can measure its length easily enough
(but one would not measure the area of the boundary).
Thus, as we will soon find, the boundary of the dragon curve has a dimension
somewhere between one and two.
\layout Standard
First one must define an iterated function system.
If in a metric space
\begin_inset Formula $S$
\end_inset
we have a list of functions
\begin_inset Formula $(f_{1},f_{2},\cdots,f_{n})$
\end_inset
, associated with a ratio list
\begin_inset Formula $(r_{1},r_{2},\cdots,r_{n})$
\end_inset
, with
\begin_inset Formula $f_{i}:S\rightarrow S$
\end_inset
, and each
\begin_inset Formula $f_{i}$
\end_inset
has a similarity ratio
\begin_inset Formula $r_{i}$
\end_inset
, a non-empty set
\begin_inset Formula $K\subseteq S$
\end_inset
is an invariant set for the iterated function system iff
\begin_inset Formula $K=f_{1}[K]\cup f_{2}[K]\cup\cdots\cup f_{n}[K]$
\end_inset
.
These iterated function systems can be represented using Maudlin-Williams
graphs, which show how the functions act, and the edges are labeled with
the similarity ratio.
The graph is made up of a set of vertices
\begin_inset Formula $V$
\end_inset
, with directed edges
\begin_inset Formula $E$
\end_inset
joining pairs of vertices.
Each function is associated with each edge, and the edge is labeled by
the appropriate ratio.
\layout Standard
For example, a square's graph would look like figure
\begin_inset LatexCommand \ref{fig:Mauldin-Williams-square}
\end_inset
.
It is made up of four functions, each of which produce a square
\begin_inset Formula $\frac{1}{2}$
\end_inset
the size of the original.
For the dragon curve, it has two functions, each of which is made up of
a dragon curve
\begin_inset Formula $\frac{1}{\sqrt{2}}$
\end_inset
the size of the original.
\layout Standard
\begin_inset Float figure
wide false
collapsed true
\layout Standard
\begin_inset Graphics
filename figures/simple_Mauldin-Williams.fig
\end_inset
\layout Caption
\begin_inset LatexCommand \label{fig:Mauldin-Williams-square}
\end_inset
Maudlin-Williams graph for a square
\end_inset
\layout Subsubsection
Similarity Dimension
\layout Standard
One of the simplest dimensions to understand is the similarity dimension.
It is defined as the (unique) number
\begin_inset Formula $s>0$
\end_inset
such that
\begin_inset Formula $\sum_{i=1}^{n}r_{i}^{s}=1$
\end_inset
is satisfied.
\layout Standard
For the square we need to solve
\begin_inset Formula $4(\frac{1}{2})^{s}=1$
\end_inset
, which gives
\begin_inset Formula $s=\frac{\log4}{\log2}=2$
\end_inset
, as expected for a two-dimensional shape.
For the (interior) of the dragon fractal we need to solve
\begin_inset Formula $2(\frac{1}{\sqrt{2}})^{s}=1$
\end_inset
, which gives
\begin_inset Formula $s=\frac{\log2}{\log\sqrt{2}}=2$
\end_inset
.
This agrees with section
\begin_inset LatexCommand \ref{sub:Space-filling}
\end_inset
.
\layout Subsubsection
Boundary of Dragon Curve
\layout Standard
\begin_inset LatexCommand \label{sub:Boundary-of-Dragon}
\end_inset
While the interior of the dragon curve is an area, the boundary of the curve
is a fractal.
It is drawn using an iterated function system.
Figure
\begin_inset LatexCommand \ref{fig:Fractle-Boundary}
\end_inset
shows how the system will have to work.
\layout Standard
\begin_inset Float figure
wide false
collapsed true
\layout Standard
\begin_inset Graphics
filename figures/fractal_boundary.eps
\end_inset
\layout Caption
\begin_inset LatexCommand \label{fig:Fractle-Boundary}
\end_inset
Fractal Boundary
\end_inset
\layout Standard
The curve is made up of two parts:
\begin_inset Formula $A$
\end_inset
and
\begin_inset Formula $B$
\end_inset
.
\begin_inset Formula $A$
\end_inset
is made up of an
\begin_inset Formula $A$
\end_inset
and a
\begin_inset Formula $B$
\end_inset
, each shrunk by a factor of
\begin_inset Formula $\sqrt{2}$
\end_inset
.
\begin_inset Formula $B$
\end_inset
is made up of two copies of
\begin_inset Formula $A$
\end_inset
, each shrunk by a factor of
\begin_inset Formula $2$
\end_inset
, but with one
\begin_inset Formula $A$
\end_inset
turned upside down.
The iterated function system is defined by four functions
\begin_inset Formula $f_{1}:A\rightarrow A$
\end_inset
,
\begin_inset Formula $f_{2}:A\rightarrow B$
\end_inset
,
\begin_inset Formula $f_{3}:B\rightarrow A$
\end_inset
and
\begin_inset Formula $f_{4}:B\rightarrow A$
\end_inset
.
This is summarised in the Maudlin-Williams graph as shown in figure
\begin_inset LatexCommand \ref{graph:Mauldin-Williams-Graph}
\end_inset
.
Note that each edge is labeled with the factor by which
\begin_inset Formula $f_{i}$
\end_inset
scales by.
\layout Standard
\begin_inset Float figure
wide false
collapsed true
\layout Standard
\begin_inset Graphics
filename figures/Mauldin-Williams.fig
\end_inset
\layout Caption
\begin_inset LatexCommand \label{graph:Mauldin-Williams-Graph}
\end_inset
Maudlin-Williams Graph
\end_inset
\layout Standard
So, the similarity dimension is found found by solving
\begin_inset Formula $1=2\left(\frac{1}{\sqrt{2}}\right)^{s}+2\left(\frac{1}{2}\right)^{s}$
\end_inset
.
We can do this if we let
\begin_inset Formula $x=\left(\frac{1}{\sqrt{2}}\right)^{s}$
\end_inset
, and solving for positive
\begin_inset Formula $x$
\end_inset
.
Then the equation becomes
\begin_inset Formula $\frac{1}{2}=x+x^{2}$
\end_inset
, which gives
\begin_inset Formula $x=\frac{\sqrt{3}-1}{2}$
\end_inset
and so
\begin_inset Formula $s=-\frac{\log x}{\log\sqrt{2}}=2.900$
\end_inset
.
\layout Subsubsection
Perron Number
\layout Standard
To find the Hausdorff dimension of the boundary of the dragon curve, first
we must calculate the Perron number.
It is defined as follows: if
\begin_inset Formula $s$
\end_inset
is a positive real number, then
\begin_inset Formula $s-$
\end_inset
dimensional Perron numbers for the graph are positive numbers
\begin_inset Formula $q_{v}$
\end_inset
(for each vertice
\begin_inset Formula $v$
\end_inset
in the graph) such that
\begin_inset Formula \[
q_{u}^{s}=\sum_{\begin{array}{c}
v\in V\\
e\in E_{uv}\end{array}}r_{e}^{s}q_{v}^{s}\]
\end_inset
\layout Standard
Thus for our square we need to solve
\begin_inset Formula $q^{s}=4\times(\frac{1}{2}\times q)^{s}$
\end_inset
.
If we let
\begin_inset Formula $q=1$
\end_inset
, we have
\begin_inset Formula $1=\frac{4}{2^{s}}$
\end_inset
which is solved to give dimension
\begin_inset Formula $s=\frac{\ln4}{\ln2}=2$
\end_inset
.
For the dragon curve boundary we need to solve the equations
\begin_inset Formula \[
q_{A}^{s}=\left(\frac{1}{\sqrt{2}}\right)^{s}q_{A}^{s}+\left(\frac{1}{\sqrt{2}}\right)^{s}q_{B}^{s}\]
\end_inset
\begin_inset Formula \[
q_{B}^{s}=2\left(\frac{1}{2}\right)^{s}q_{A}^{s}\]
\end_inset
\layout Standard
This is done, like for the similarity dimension, by letting
\begin_inset Formula $\lambda=\sqrt{2}^{s}$
\end_inset
,
\begin_inset Formula $x=q_{A}^{s}$
\end_inset
and
\begin_inset Formula $y=q_{B}^{s}$
\end_inset
.
Then the equations become
\begin_inset Formula $x=\frac{x+y}{\lambda}$
\end_inset
,
\begin_inset Formula $y=2\frac{x}{\lambda^{2}}$
\end_inset
.
If we let
\begin_inset Formula $y=1$
\end_inset
we have
\begin_inset Formula $x(1-\lambda)+1=0$
\end_inset
and
\begin_inset Formula $2x=\lambda^{2}$
\end_inset
.
Combining these gives
\begin_inset Formula $\lambda^{2}(1-\lambda)+2=0\Rightarrow\lambda^{3}-\lambda^{2}-2=0$
\end_inset
.
The positive root of this equation
\begin_inset Foot
collapsed true
\layout Standard
Calculated using the Newtonian method.
That is the roots of
\begin_inset Formula $f(x)=0$
\end_inset
can be found by setting
\begin_inset Formula $x_{n+1}=x_{n}-\frac{f(x)}{f'(x)}$
\end_inset
.
\end_inset
is
\begin_inset Formula $\lambda=1.69562$
\end_inset
.
So
\begin_inset Formula $s=\frac{\ln\lambda}{\ln\sqrt{2}}=1.52363\approx1.52$
\end_inset
.
This is between 1 and 2 as expected - the length of the boundary is infinite,
but it has zero area, so the dimension must be
\begin_inset Formula $\geq1$
\end_inset
but
\begin_inset Formula $\leq2$
\end_inset
\layout Subsubsection
Hausdorff Dimension
\layout Standard
The Hausdorff dimension is defined as follows: If
\begin_inset Formula $U$
\end_inset
is a non-empty subset of
\begin_inset Formula $\mathbb{R}^{n}$
\end_inset
, the diameter of
\begin_inset Formula $U$
\end_inset
is defined as
\begin_inset Formula $\left|U\right|=\sup\left\{ \left|x-y\right|:x,y\in U\right\} $
\end_inset
, that is the furthest distance between any two points in
\begin_inset Formula $U$
\end_inset
.
If
\begin_inset Formula $\{ U_{i}\}$
\end_inset
is a countable collection of sets of diameter
\begin_inset Formula $|U_{i}|\leq\delta$
\end_inset
that cover
\begin_inset Formula $F$
\end_inset
(that is
\begin_inset Formula $F\subset\bigcup_{i}U_{i}$
\end_inset
) and we say that
\begin_inset Formula $\{ U_{i}\}$
\end_inset
is a
\begin_inset Formula $\delta$
\end_inset
-cover of
\begin_inset Formula $F$
\end_inset
.
Suppose
\begin_inset Formula $F\subset\mathbb{R}^{n}$
\end_inset
and
\begin_inset Formula $s$
\end_inset
a non-negative number.
For any
\begin_inset Formula $\delta>0$
\end_inset
with
\begin_inset Formula $\left\{ U_{i}\right\} $
\end_inset
as a
\begin_inset Formula $\delta$
\end_inset
-cover for
\begin_inset Formula $F$
\end_inset
define
\begin_inset Formula \[
\mathcal{H}_{\delta}^{s}(F)=\inf\left\{ \sum_{i}\left|U_{i}\right|^{s}\right\} \]
\end_inset
\layout Standard
Then take the limit
\begin_inset Formula $\mathcal{H}^{s}(F)=\lim_{\delta\rightarrow0}\mathcal{H}_{\delta}^{s}(F)$
\end_inset
.
The Hausdorff dimension
\begin_inset Formula $s$
\end_inset
is the value for for which
\begin_inset Formula \[
\mathcal{H}^{s'}(F)=\{\begin{array}{ccc}
\infty & \, & s'>s\\
0 & \, & s'
|
\begin_inset Text
\layout Standard
\begin_inset Formula $\left[c_{0},\overrightarrow{w},x-\frac{q_{n-1}}{q_{n}}\right]$
\end_inset
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $\leftrightarrow\left(\begin{array}{cc}
p_{n} & p_{n-1}\\
q_{n} & q_{n-1}\end{array}\right)\left(\begin{array}{cc}
x-\frac{q_{n-1}}{q_{n}} & 1\\
1 & 0\end{array}\right)$
\end_inset
\end_inset
|
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $=\left(\begin{array}{cc}
xp_{n}-\left(p_{n}q_{n-1}-p_{n-1}q_{n}\right)/q_{n} & p_{n}\\
xq_{n} & q_{n}\end{array}\right)$
\end_inset
\end_inset
|
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $=\left(\begin{array}{cc}
xp_{n}-\left(-1\right)^{n+1}/q_{n} & p_{n}\\
xq_{n} & q_{n}\end{array}\right)$
\end_inset
\end_inset
|
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $\leftrightarrow\frac{xp_{n}+\left(-1\right)^{n}/q_{n}}{xq_{n}}$
\end_inset
\end_inset
|
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $=\frac{p_{n}}{q_{n}}+\frac{\left(-1\right)^{n}}{xq_{n}^{2}}$
\end_inset
\end_inset
|
\end_inset
\layout Standard
But
\begin_inset Formula $x-\frac{q_{n-1}}{q_{n}}=x-\frac{1}{q_{n}/q_{n-1}}=\left[x,-\overleftarrow{w}\right].$
\end_inset
So
\begin_inset Formula \[
\left[c_{0},\overrightarrow{w},x-\frac{q_{n-1}}{q_{n}}\right]=\left[c_{0},\overrightarrow{w},x,-\overleftarrow{w}\right]=\frac{p_{n}}{q_{n}}+\frac{\left(-1\right)^{n}}{xq_{n}^{2}}\]
\end_inset
\layout Standard
Note that
\begin_inset Formula $\overrightarrow{w},x,-\overleftarrow{w}$
\end_inset
reminds us of our folding operation (see section
\begin_inset LatexCommand \ref{sub:algoritm2}
\end_inset
).
\layout Subsection
Paperfolding
\layout Standard
It is not generally easy to find a continued fraction representation of
any number, but there are some (infinitely many, but not all of
\begin_inset Formula $\mathbb{R}$
\end_inset
) that, with the help of our beloved paperfolding sequence, we can.
Proposition, if we have the paperfolding instructions
\begin_inset Formula $\epsilon_{0}\epsilon_{1}\epsilon_{2}\cdots$
\end_inset
(
\begin_inset Formula $\epsilon_{k}\in\{-1,+1\}$
\end_inset
), then any number of the form
\begin_inset Formula $x\sum_{k=0}^{n}\epsilon_{k}x^{-2^{k}}$
\end_inset
has the continued fraction expansion
\begin_inset Formula \[
x\sum_{k=0}^{n}\epsilon_{k}x^{-2^{k}}=\left[\epsilon_{0},T_{-\epsilon_{n}x}\cdots T_{-\epsilon_{3}x}\circ T_{-\epsilon_{2}x}(\epsilon_{1}x)\right]\]
\end_inset
\layout Standard
For
\begin_inset Formula $n=1,3,4,\cdots$
\end_inset
and where
\begin_inset Formula $T_{i}$
\end_inset
is similar to our folding operation defined in section
\begin_inset LatexCommand \ref{sub:algoritm2}
\end_inset
, and is defined as
\begin_inset Formula $T_{\overrightarrow{p}}(\overrightarrow{w})=\overrightarrow{w},\overrightarrow{p},-\overrightarrow{w}$
\end_inset
.
For example
\begin_inset Formula $T_{-\epsilon_{2}x}(\epsilon_{1}x)=\epsilon_{1}x,-\epsilon_{2}x,-\epsilon_{1}x$
\end_inset
.
\layout Standard
Our proof is by induction.
First we show it is true for
\begin_inset Formula $n=1$
\end_inset
.
For LHS we have
\begin_inset Formula \[
x\sum_{k=0}^{1}\epsilon_{k}x^{-2^{k}}=\epsilon_{0}+\frac{\epsilon_{1}}{x}=[\epsilon_{0},\frac{x}{\epsilon_{1}}]=[\epsilon_{0},\epsilon_{1}x]\]
\end_inset
\layout Standard
(
\begin_inset Formula $1/\epsilon_{k}=\epsilon_{k}$
\end_inset
as
\begin_inset Formula $\epsilon_{k}\in\{-1,+1\}$
\end_inset
).
RHS is
\begin_inset Formula $\left[\epsilon_{0},\epsilon_{1}x\right]$
\end_inset
so our proposition is true for
\begin_inset Formula $n=1$
\end_inset
.
Also
\begin_inset Formula $[\epsilon_{0},\epsilon_{1}x]=p_{1}/q_{1}$
\end_inset
(
\begin_inset Formula $p_{i}$
\end_inset
,
\begin_inset Formula $q_{i}$
\end_inset
\begin_inset Formula $\in\mathbb{Z}$
\end_inset
).
Let
\begin_inset Formula $q_{1}=x$
\end_inset
.
Now, we assume it is true for a
\begin_inset Formula $n$
\end_inset
, that is:
\layout Standard
\begin_inset Formula \[
\epsilon_{0}+\frac{\epsilon_{1}}{x}+\frac{\epsilon_{2}}{x^{3}}+\dots+\frac{\epsilon_{n}}{x^{2^{n}-1}}=[1,\overrightarrow{w}]=\frac{p_{n}}{q_{n}}\]
\end_inset
\layout Standard
Since
\begin_inset Formula $\left|\overrightarrow{w}\right|=n$
\end_inset
is odd,
\begin_inset Formula $(-1)^{n}=-1$
\end_inset
(see section
\begin_inset LatexCommand \ref{sub:Total-Number-of}
\end_inset
).
As the biggest denominator is
\begin_inset Formula $x^{2^{n}-1}$
\end_inset
, we can take
\begin_inset Formula $q_{n}=x^{2^{n}-1}$
\end_inset
.
We will want the
\begin_inset Formula $q_{n+1}$
\end_inset
to be
\begin_inset Formula $x^{2^{n+1}-1}=xx^{2^{n+1}-2}=xx^{2(2^{n}-1)}=xq_{n}^{2}$
\end_inset
.
So for
\begin_inset Formula $n+1$
\end_inset
we have, using the property in section
\begin_inset LatexCommand \ref{sub:Properties}
\end_inset
,
\layout Standard
\begin_inset Tabular
|
\begin_inset Text
\layout Standard
\begin_inset Formula $\epsilon_{0}+\frac{\epsilon_{1}}{x}+\dots+\frac{\epsilon_{n}}{x^{2^{n}-1}}+\frac{\epsilon_{n+1}}{x^{2^{n+1}-1}}$
\end_inset
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $=\frac{p_{n}}{q_{n}}+\frac{\epsilon_{n+1}}{xq_{n}^{2}}$
\end_inset
\end_inset
|
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $=\frac{p_{n}}{q_{n}}+\frac{(-1)^{n}}{-\epsilon_{n+1}xq_{n}^{2}}$
\end_inset
\end_inset
|
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $=[1,\overrightarrow{w},-\epsilon_{n+1}x,-\overleftarrow{w}]$
\end_inset
\end_inset
|
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $=[1,T_{-\epsilon_{n+1}x}(\overrightarrow{w})]$
\end_inset
\end_inset
|
\end_inset
\layout Standard
So we have proved our proposition by induction.
\layout Section
Quasi-Crystals
\layout Standard
Before beginning the analysis of one dimensional crystals we will first
look at the definitions of periodic and aperiodic sequences, before looking
at how they are related to crystals.
\layout Subsection
Periodic Sequences
\layout Standard
A [complex] function
\begin_inset Formula $f(x)$
\end_inset
is said to be periodic iff
\begin_inset Formula \[
f(x+n\tau)=f(x)\]
\end_inset
\layout Standard
For some
\begin_inset Formula $n\in\mathbb{Z}$
\end_inset
.
\begin_inset Formula $\tau$
\end_inset
is said to be the period of
\begin_inset Formula $f(x)$
\end_inset
(note we choose the lowest such value of
\begin_inset Formula $\tau$
\end_inset
as any integer multiple of
\begin_inset Formula $\tau$
\end_inset
would also satisfy the condition).
\layout Subsection
Fourier Transform and Series
\layout Standard
The [one dimensional] Fourier transform is defined by
\begin_inset Formula \[
F(u)=\mathcal{F}(f(x),u)=\int_{-\infty}^{\infty}f(x)e^{-i2\pi ux}dx\]
\end_inset
\layout Standard
It maps a function in normal space to one in Fourier, or reciprocal, space.
A single point in the function will correspond to a wave with a frequency
of that point in wave space, and a wave in the function will be a single
point in wave space.
The proof of the statement follows:
\layout Standard
If
\begin_inset Formula $f(x)$
\end_inset
is a single point at
\begin_inset Formula $x_{0}$
\end_inset
, we have
\begin_inset Formula $f(x)=\delta(x-x_{0})$
\end_inset
.
Then
\begin_inset Formula $F(u)=\int_{-\infty}^{\infty}\delta(x-x_{0})e^{-i2\pi ux}dx=e^{-2\pi ix_{0}u}$
\end_inset
, that is a wave of frequency
\begin_inset Formula $x_{0}$
\end_inset
.
Conversely, if
\begin_inset Formula $f(x)$
\end_inset
is a wave, that is
\begin_inset Formula $f(x)=e^{2\pi iu_{0}x}$
\end_inset
we have
\begin_inset Formula $F(u)=\int_{-\infty}^{\infty}e^{2\pi i(u_{0}-u)x}dx=\delta(u_{0}-u)$
\end_inset
.
\layout Standard
Periodic function can be expressed as a Fourier series
\begin_inset Formula \[
f(x)=\sum_{k=0}^{\infty}c_{k}e^{ikx}\]
\end_inset
\layout Standard
where
\begin_inset Formula $c_{k}\in\mathbb{C}$
\end_inset
,are unique, and can be calculated as
\begin_inset Formula $c_{k}=\frac{1}{\tau}\int_{0}^{\tau}f(x)e^{-ikx}dx$
\end_inset
due to orthogonality:
\begin_inset Formula $\int_{0}^{\tau}e^{inx}e^{-mx}dx=\tau\delta_{n,m}$
\end_inset
\begin_inset Foot
collapsed true
\layout Standard
\begin_inset Formula $\int_{0}^{\tau}e^{inx}e^{-mx}dx=\int_{0}^{\tau}e^{i(n-m)x}dx=[\frac{e^{i(m-n)x}}{i(m-n)}]_{0}^{\tau}=0$
\end_inset
if
\begin_inset Formula $m\neq n$
\end_inset
, but if
\begin_inset Formula $m=n$
\end_inset
we get
\begin_inset Formula $\int_{0}^{\tau}e^{i(n-m)x}dx=\int_{0}^{\tau}e^{0}dx=\int_{0}^{\tau}1dx=\tau$
\end_inset
.
\end_inset
.
Note, we could choose a scale for
\begin_inset Formula $x$
\end_inset
such that
\begin_inset Formula $\tau=2\pi$
\end_inset
.
Fundamentally, the series is a sum of waves of period
\begin_inset Formula $\tau$
\end_inset
and its harmonies.
As a periodic function is made up of a sum of distinct frequencies, its
Fourier transform will be a set of distinct points.
As all the frequencies (of which there are countably many) are harmonies
on one base frequency, the points in wave space will display certain symmetries.
\layout Subsection
Aperiodic Sequences
\layout Standard
If our function is over a metric space, we can define a distance between
two function
\begin_inset Formula $\rho(f(x),g(x))$
\end_inset
.
The metric will be defined as
\begin_inset Formula $\rho(f(x),g(x))=\sup_{x}\left|f(x)-g(x)\right|$
\end_inset
.
We will consider functions defined by finite sums
\begin_inset Formula \[
f(x)=\sum_{k=0}^{N}c_{k}e^{i\lambda_{k}x}\]
\end_inset
\layout Standard
where
\begin_inset Formula $\lambda_{k}\in\mathbb{R}$
\end_inset
.
In the special case that the set
\begin_inset Formula $\{\lambda_{k}\}$
\end_inset
correspond to a single frequency and its harmonies, we one again have the
Fourier series, and so a periodic function.
First we shall consider the functions
\begin_inset Formula $f(x)=e^{i\lambda_{1}x}+e^{i\lambda_{2}x}$
\end_inset
.
If
\begin_inset Formula $\frac{\lambda_{1}}{\lambda_{2}}$
\end_inset
is rational, then we have the periodic Fourier case again, but if it is
irrational then the function is not periodic, but it can be shown that
there are arbitrarily great integers
\begin_inset Formula $n_{1}$
\end_inset
,
\begin_inset Formula $n_{2}$
\end_inset
such that
\begin_inset Formula $\left|\frac{n_{1}}{\lambda_{1}}-\frac{n_{2}}{\lambda_{2}}\right|<\epsilon$
\end_inset
for any given
\begin_inset Formula $\epsilon>0$
\end_inset
.
Now, if we let
\begin_inset Formula $\tau$
\end_inset
be a number which is
\emph on
near
\emph default
(say between)
\begin_inset Formula $\frac{n_{1}}{\lambda_{1}}$
\end_inset
and
\begin_inset Formula $\frac{n_{2}}{\lambda_{2}}$
\end_inset
then
\begin_inset Formula $\tau$
\end_inset
is
\emph on
almost
\emph default
a period of
\begin_inset Formula $e^{i\lambda_{1}x}$
\end_inset
and
\begin_inset Formula $e^{i\lambda_{2}x}$
\end_inset
.
Thus the difference
\begin_inset Formula $f(x+\tau)-f(x)$
\end_inset
is very small
\begin_inset Formula $\forall x$
\end_inset
.
Note that for a periodic function we can choose
\begin_inset Formula $\tau$
\end_inset
so that this difference is zero.
A real number
\begin_inset Formula $\tau$
\end_inset
is called a
\emph on
translation number
\emph default
of
\begin_inset Formula $f(x)$
\end_inset
if
\begin_inset Formula $\left|f(x+\tau)-f(x)\right|\leq\epsilon$
\end_inset
for a given
\begin_inset Formula $\epsilon>0$
\end_inset
.
\begin_inset Formula $\tau$
\end_inset
is determined by
\begin_inset Formula $\epsilon$
\end_inset
, so
\begin_inset Formula $\tau=\tau(\epsilon)$
\end_inset
.
It can be proved that
\begin_inset Formula $\tau(\epsilon_{1})+\tau(\epsilon_{2})=\tau(\epsilon_{1}+\epsilon_{2})$
\end_inset
,
\begin_inset Formula $\tau=-\tau$
\end_inset
and so
\begin_inset Formula $\tau(\epsilon_{1})-\tau(\epsilon_{2})=\tau(\epsilon_{1}+\epsilon_{2})$
\end_inset
\begin_inset Foot
collapsed true
\layout Standard
See reference
\begin_inset LatexCommand \cite{aperiodic}
\end_inset
for details and proofs.
\end_inset
.
\layout Standard
To define an aperiodic function we need one more element.
For each
\begin_inset Formula $\epsilon$
\end_inset
there a set of real translation numbers
\begin_inset Formula $T$
\end_inset
.
In order for the functions described by translation number to remain closed
under sensible things like addition, we require that
\begin_inset Formula $T$
\end_inset
be
\emph on
relatively dense
\emph default
.
That is, a length
\begin_inset Formula $L$
\end_inset
exists (for each
\begin_inset Formula $\epsilon$
\end_inset
) such that any interval of that length from the real line contains at least
one translation number [from
\begin_inset Formula $T$
\end_inset
].
\layout Standard
For example, for a periodic function of period
\begin_inset Formula $\tau$
\end_inset
, the translation numbers can be taken to be
\begin_inset Formula $E=\{\lambda_{k}\}=\{ k\tau\}$
\end_inset
.
Then the length
\begin_inset Formula $L$
\end_inset
can be taken to be any number greater than
\begin_inset Formula $\tau$
\end_inset
, so
\begin_inset Formula $E$
\end_inset
is relatively dense.
\layout Standard
The definition: A continuous function
\begin_inset Formula $f(x)$
\end_inset
is aperiodic iff, for any given
\begin_inset Formula $\epsilon>0$
\end_inset
, a relatively dense set of translation number of
\begin_inset Formula $f(x)$
\end_inset
(corresponding to
\begin_inset Formula $\epsilon$
\end_inset
) exists.
\layout Standard
Note, periodic function are a [special] subset of aperiodic functions, and
that its Fourier transform will be a discrete set of
\begin_inset Formula $N$
\end_inset
points corresponding to the frequencies
\begin_inset Formula $\lambda$
\end_inset
.
Thus it is understanding the Fourier transform of a function that is crucial
in determining if a function is aperiodic/periodic or not.
\layout Subsection
Lattice
\layout Standard
\begin_inset LatexCommand \label{sub:Lattice}
\end_inset
A lattice is a set of points in space.
It is a discrete set of points, determined by a finite set of basis vectors.
Every point in the lattice can be represented by a linear combination of
these vectors, with whole number coefficients.
For example, if in two dimensional space we had the basis vectors
\begin_inset Formula $\left(\begin{array}{c}
1\\
0\end{array}\right)$
\end_inset
and
\begin_inset Formula $\left(\begin{array}{c}
0\\
1\end{array}\right)$
\end_inset
, this would generate the lattice shown in figure
\begin_inset LatexCommand \ref{fig:lattices}
\end_inset
.
The other lattice was generated with basis vectors
\begin_inset Formula $\left\{ \left(\begin{array}{c}
1\\
0\end{array}\right),\left(\begin{array}{c}
1/2\\
1\end{array}\right)\right\} $
\end_inset
.
\layout Standard
associated
\begin_inset Float figure
placement h
wide false
collapsed true
\layout Standard
\begin_inset Tabular
|
\begin_inset Text
\layout Standard
\begin_inset Graphics
filename figures/lattice1.fig
lyxscale 50
scale 50
\end_inset
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Graphics
filename figures/lattice3.fig
lyxscale 50
scale 50
\end_inset
\end_inset
|
|
\begin_inset Text
\layout Standard
\begin_inset Formula $\left\{ \left(\begin{array}{c}
1\\
0\end{array}\right),\left(\begin{array}{c}
0\\
1\end{array}\right)\right\} $
\end_inset
\end_inset
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $\left\{ \left(\begin{array}{c}
1\\
0\end{array}\right),\left(\begin{array}{c}
1/2\\
1\end{array}\right)\right\} $
\end_inset
\end_inset
|
\end_inset
\layout Caption
\begin_inset LatexCommand \label{fig:lattices}
\end_inset
Lattices
\end_inset
\layout Subsection
Delone Sets
\layout Standard
A set of points
\begin_inset Formula $\Lambda$
\end_inset
in euclidean space is a
\emph on
Delone set
\emph default
iff it is
\emph on
discrete
\emph default
and
\emph on
relatively dense
\emph default
.
Here discrete means that the points in the set cannot be arbitrarily close
together:
\begin_inset Formula $\exists$
\end_inset
a radius
\begin_inset Formula $r>0$
\end_inset
such that for every
\begin_inset Formula $x,y\in\Lambda$
\end_inset
,
\begin_inset Formula $\left|x-y\right|\geq r$
\end_inset
.
Relatively dense in euclidean space means that the points cover the space:
\begin_inset Formula $\exists$
\end_inset
a radius
\begin_inset Formula $R>0$
\end_inset
such that every sphere (or circle etc) of radius
\begin_inset Formula $\geq R$
\end_inset
in space contain at least one point in
\begin_inset Formula $\Lambda$
\end_inset
.
\layout Standard
For example, in
\begin_inset Formula $\mathbb{R}$
\end_inset
, the set
\begin_inset Formula $\Lambda=\{\frac{1}{n},n:n\in\mathbb{Z}\}$
\end_inset
is relatively dense as any line segment of size
\begin_inset Formula $1$
\end_inset
or greater in
\begin_inset Formula $\mathbb{R}$
\end_inset
contains at least one point of
\begin_inset Formula $\Lambda$
\end_inset
.
However, since the points
\begin_inset Formula $\frac{1}{n}$
\end_inset
converge to zero, we can alway find an
\begin_inset Formula $\epsilon$
\end_inset
such that
\begin_inset Formula $\left|x-y\right|<\epsilon$
\end_inset
, therefore the set is not discrete, and thus not a Delone set.
The set
\begin_inset Formula $\Lambda\{\pm n^{2}:n\in\mathbb{Z}\}$
\end_inset
is discrete (let
\begin_inset Formula $r=1$
\end_inset
), but it is not relatively dense as, for any finite length, there will
always be a line line segment that does not include
\begin_inset Formula $\Lambda$
\end_inset
as
\begin_inset Formula $(n+1)^{2}-n^{2}\rightarrow\infty$
\end_inset
.
\layout Standard
A normal crystal lattice is always a Delone set.
This follows from the construction of the lattice from a set of basis vectors:
the distance between any two points is no less than the size of the smallest
basis vector etc.
\layout Standard
The
\begin_inset Formula $r$
\end_inset
-star (
\begin_inset Formula $r>0$
\end_inset
) is the point set
\begin_inset Formula $\overline{B}(x,r)\cap\Lambda$
\end_inset
where
\begin_inset Formula $x\in\Lambda$
\end_inset
and
\begin_inset Formula $\overline{B}(x,r)$
\end_inset
is the closed ball of radius
\begin_inset Formula $r$
\end_inset
centered at
\begin_inset Formula $x$
\end_inset
.
The
\begin_inset Formula $r$
\end_inset
-star will contain only a finite number of points as
\begin_inset Formula $\Lambda$
\end_inset
is discrete.
\begin_inset Formula $\Lambda$
\end_inset
is a
\emph on
regular set
\emph default
iff the
\begin_inset Formula $r$
\end_inset
-stars of
\begin_inset Formula $\Lambda$
\end_inset
are congruent for every
\begin_inset Formula $r>0$
\end_inset
(congruency means that it is identical under translation, rotations, and
reflections).
Thus the crystal lattice is a regular, or periodic, Delone set.
\layout Subsection
Voronoi Cells
\layout Standard
The crystal lattices can be thought of as being made up from identical
\emph on
cells
\emph default
, which are tiled together to form the lattice.
The
\emph on
unit cell
\emph default
is defined to have edges which are parallel to the basis (see section
\begin_inset LatexCommand \ref{sub:Lattice}
\end_inset
).
The unit cell will depend on the choice of basis used, and not all choices
of cell will demonstrate the symmetry of the full lattice.
Figure
\begin_inset LatexCommand \ref{cap:Unit-Cell}
\end_inset
shows two possible unit cells for the same lattice.
\layout Standard
\begin_inset Float figure
placement h
wide false
collapsed true
\layout Standard
\begin_inset Graphics
filename figures/unit_cell.fig
\end_inset
\layout Caption
\begin_inset LatexCommand \label{cap:Unit-Cell}
\end_inset
Unit Cell
\end_inset
\layout Standard
The
\emph on
Voroni cell
\emph default
, on the other hand,will always display the symmetry of the lattice, and
is independent of the chosen basis.
It is defined to be the set of points centered about a lattice point such
that they are closer to the center lattice point than to any other lattice
point.
Formally, this can be written
\begin_inset Formula $\forall x\in\Lambda$
\end_inset
as
\begin_inset Formula \[
V(x)=\left\{ u\in\mathbb{R}^{n}\;:\;\left|x-u\right|\leq\left|y-u\right|\;\forall y\in\Lambda\right\} \]
\end_inset
\layout Standard
It is constructed by drawing lines between all the lattice points and the
center one.
These lines are then bisected at right angles by drawing other lines (lines
in two dimensions, planes in three dimension etc.).
The Voroni cell is the smallest area (or volume etc.) bounded by the other
lines (or planes etc.).
See figure
\begin_inset LatexCommand \ref{cap:Voronoi-Cell}
\end_inset
for an example.
\layout Standard
\begin_inset Float figure
placement h
wide false
collapsed true
\layout Standard
\begin_inset Graphics
filename figures/voronoi_cell.fig
\end_inset
\layout Caption
\begin_inset LatexCommand \label{cap:Voronoi-Cell}
\end_inset
Voronoi Cell
\end_inset
\layout Subsection
Isometries
\layout Standard
An isometry is a transformation that preserves distance and between points
(and angles between lines).
Isometries in two dimensions can be translations, rotations and reflections.
Reflections and rotations also fix the origin.
\layout Standard
With an appropriate basis, the isometries that fix the origin (these are
called linear isometries) can be represented by a
\begin_inset Formula $n\times n$
\end_inset
matrix (in
\begin_inset Formula $n$
\end_inset
-dimensional space) of the form
\begin_inset Formula \[
\left(\begin{array}{cccc}
A_{1} & 0 & 0 & 0\\
0 & A_{2} & 0 & 0\\
0 & 0 & \ddots & 0\\
0 & 0 & 0 & A_{k}\end{array}\right)\]
\end_inset
\layout Standard
where each
\begin_inset Formula $A_{i}$
\end_inset
is either
\begin_inset Formula $1$
\end_inset
,
\begin_inset Formula $-1$
\end_inset
or
\begin_inset Formula $\left(\begin{array}{cc}
\cos\theta & -\sin\theta\\
\sin\theta & \cos\theta\end{array}\right)$
\end_inset
(which rotates by an angle
\begin_inset Formula $\theta$
\end_inset
).
From the definition of lattice, with respect to that basis the points must
have integer co-ordinates, so our linear isometries can be represented
by a (finite) subgroup of
\begin_inset Formula $\textrm{GL}(n,\mathbb{Z})$
\end_inset
, the general linear group of matrices of degree
\begin_inset Formula $n$
\end_inset
with integer elements and determinant
\begin_inset Formula $\pm1$
\end_inset
.
\layout Subsubsection
The Crystallographic Restriction
\layout Standard
\begin_inset LatexCommand \label{sub:restriction}
\end_inset
Now, any linear isometry acting on our crystal lattice can, depending on
the basis chosen, be represented by either the former of latter form of
matrix.
As these matrices are the same but for a change of basis, they are conjugate,
and thus have the same trace.
For
\begin_inset Formula $2$
\end_inset
-dimensions, the
\begin_inset Formula $\textrm{trace}(\textrm{GL}(n,\mathbb{Z}))\in\mathbb{Z}$
\end_inset
, but the
\begin_inset Formula $\textrm{trace}\left(\begin{array}{cc}
\cos\theta & -\sin\theta\\
\sin\theta & \cos\theta\end{array}\right)=2\cos\theta$
\end_inset
.
As
\begin_inset Formula $\left|\cos\theta\right|\leq1$
\end_inset
(and assume
\begin_inset Formula $0\leq\theta<2\pi$
\end_inset
), we have
\begin_inset Formula $\cos\theta=\{0,\pm\frac{1}{2},\pm1\}$
\end_inset
, so
\begin_inset Formula $\theta=\{\frac{\pi}{2},\frac{\pi}{3},2\frac{\pi}{3},0,\pi\}$
\end_inset
.
Thus the only rotational symmetry that a crystal can have is that of order
4,6,3 and 2 (that is its point group contains rotations of order 2,3,4
and 6).
There are similar restrictions for higher dimensional crystals.
\layout Standard
Point groups containing the rotations of order 5, 7, 8 etc will be called
\emph on
quasi-crystallographic
\emph default
.
\layout Subsection
Quasicrystals
\layout Standard
It was once thought that the long range order could only be due to periodicity.
However in 1984 a crystal with long range symmetry was found that did not
have a periodic structure.
Such crystals were named
\emph on
quasicrystals
\emph default
.
A quasicrystal is a peculiar form of solid in which the atoms of the solid
are arranged in a seemingly regular, yet non-repeating structure.
This kind of structure had actually been seen before in other places, in
particular Penrose tiles
\begin_inset LatexCommand \cite{quasicrystal}
\end_inset
.
\layout Standard
A periodic crystal is one that has translational symmetry (in all
\begin_inset Formula $n$
\end_inset
dimensions).
An aperiodic crystal is one which has no translational symmetry.
(Crystals which are only periodic in some of the
\begin_inset Formula $n$
\end_inset
directions, but aperiodic in the others are called subperiodic).
The quasicrystal is defined as an aperiodic crystal which exhibits a diffractio
n pattern forbidden for crystals.For example a 5-fold rotational symmetry
in its diffraction pattern is forbidden for a crystal, see section
\begin_inset LatexCommand \ref{sub:restriction}
\end_inset
.
\layout Subsection
X-Ray Diffraction
\layout Standard
Crystals are studied by looking at their x-ray diffraction patters, see
figure
\begin_inset LatexCommand \ref{fig:x-ray-diffraction}
\end_inset
.
They can be used to find Bragg peaks, and determine information about the
crystal's structure.
The diffraction of x-rays in crystals can be modelled by Fraunhofer diffraction
due to the following approximations: the light [x-ray] source is assumed
to be very far away from the subject, so the incoming light rays can be
assumed to be parallel; the detection screen is far enough away that light
leaving the subject at the same angle meet at the same point on the screen;
and only one frequency of light is used (not a spectrum).
\layout Standard
\begin_inset Float figure
placement h
wide false
collapsed true
\layout Standard
\begin_inset Graphics
filename figures/x-raydiffraction.eps
\end_inset
\layout Caption
\begin_inset LatexCommand \label{fig:x-ray-diffraction}
\end_inset
x-ray diffraction
\end_inset
\layout Standard
By observing that a disc as the subject will create a diffraction pattern
of a spot whose radius is inversely proportional to the original disc (see
figure
\begin_inset LatexCommand \ref{cap:DiffractionCircle}
\end_inset
), we can assume that the limit of a point would produce a diffraction pattern
of constant intensity over the whole screen.
If we let
\begin_inset Formula $\overrightarrow{s}$
\end_inset
be the point on the screen,
\begin_inset Formula $J(\overrightarrow{s})$
\end_inset
the intensity at that point, then we have
\begin_inset Formula $J(\overrightarrow{s})\equiv1$
\end_inset
.
Also define the amplitude
\begin_inset Formula $A(\overrightarrow{s})$
\end_inset
at that point so that
\begin_inset Formula $J(\overrightarrow{s})=\left|A(\overrightarrow{s})\right|^{2}$
\end_inset
.
Since
\begin_inset Formula $\left|A(\overrightarrow{s})\right|=1$
\end_inset
,
\begin_inset Formula $A(\overrightarrow{s})=e^{-i2\pi f(\overrightarrow{s})}$
\end_inset
where
\begin_inset Formula $f(\overrightarrow{s})$
\end_inset
is some real values function, which can be determined by considering two
points.
\layout Standard
\begin_inset Float figure
placement !h
wide false
collapsed true
\layout Standard
\begin_inset Tabular
|
\begin_inset Text
\layout Standard
\begin_inset Graphics
filename figures/circles/big_circle.pgm
scale 50
\end_inset
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Graphics
filename figures/circles/small_circle.pgm
scale 50
\end_inset
\end_inset
|
|
\begin_inset Text
\layout Standard
\begin_inset Graphics
filename figures/circles/big_circle_fourier.pgm
scale 50
\end_inset
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Graphics
filename figures/circles/small_circle_fourier.pgm
scale 50
\end_inset
\end_inset
|
|
\begin_inset Text
\layout Standard
big circle
\end_inset
|
\begin_inset Text
\layout Standard
small circle
\end_inset
|
\end_inset
\layout Caption
\begin_inset LatexCommand \label{cap:DiffractionCircle}
\end_inset
Diffraction Patterns of Circles
\end_inset
\layout Standard
Let the two points be separated by
\begin_inset Formula $\overrightarrow{d}$
\end_inset
.
From our assumptions, light wave from the two points will leave the points
in phase, but will arrive at the point
\begin_inset Formula $\overrightarrow{s}$
\end_inset
out of phase by an amount determined by the separation of the two points.
This phase will be given by
\begin_inset Formula $\overrightarrow{s}.\overrightarrow{d}$
\end_inset
.
This will cause the two light rays to constructively interfere at points
where the phase is an integer multiple of the wavelength
\begin_inset Formula $\overrightarrow{s}.\overrightarrow{d}=n\lambda$
\end_inset
, but to cancel each other at
\begin_inset Formula $\overrightarrow{s}.\overrightarrow{d}=(n+\frac{1}{2})\lambda$
\end_inset
(
\begin_inset Formula $n\in\mathbb{Z}$
\end_inset
).Thus the diffraction pattern will go from light to dark with period inversely
proportional to the distance
\begin_inset Formula $\left|\overrightarrow{d}\right|$
\end_inset
between the points.
\layout Standard
Now, the amplitude is additive, so
\begin_inset Formula $A(\overrightarrow{s})=A_{0}(\overrightarrow{s})+A_{1}(\overrightarrow{s})$
\end_inset
where
\begin_inset Formula $A_{k}(\overrightarrow{s})$
\end_inset
is the amplitude function from the respective point.
As
\begin_inset Formula $A_{1}(\overrightarrow{s})$
\end_inset
is just
\begin_inset Formula $A_{0}(\overrightarrow{s})$
\end_inset
with a phase shift of
\begin_inset Formula $\overrightarrow{s}.\overrightarrow{d}$
\end_inset
, we should let
\begin_inset Formula $A_{1}(\overrightarrow{s})=e^{2\pi i\overrightarrow{s}.\overrightarrow{d}}A_{0}(\overrightarrow{s})$
\end_inset
.
So
\begin_inset Formula $A(\overrightarrow{s})=A_{0}(\overrightarrow{s})(1+e^{2\pi i\overrightarrow{s}.\overrightarrow{d}})$
\end_inset
, and therefore
\begin_inset Formula $J(\overrightarrow{s})=\left|A(\overrightarrow{s})\right|^{2}=\left|A_{0}(\overrightarrow{s})\right|^{2}(1+e^{2\pi i\overrightarrow{s}.\overrightarrow{d}})(1+e^{-2\pi i\overrightarrow{s}.\overrightarrow{d}})=2(1+cos(2\pi\overrightarrow{s}.\overrightarrow{d}))$
\end_inset
.
\begin_inset Formula $|\overrightarrow{s}.\overrightarrow{d}|=|\overrightarrow{d}|$
\end_inset
so the period of the intensity is inversely proportional to the distance
between the two points.
\layout Standard
For an arbitrary number of points,
\begin_inset Formula $N$
\end_inset
say, we have
\begin_inset Formula $A(\overrightarrow{s})=A_{0}(\overrightarrow{s})+A_{1}(\overrightarrow{s})+\cdots A_{N-1}(\overrightarrow{s})=A_{0}(\overrightarrow{s})\sum_{k=0}^{N-1}e^{2\pi i\overrightarrow{s}.\overrightarrow{d_{k}}}$
\end_inset
where
\begin_inset Formula $\overrightarrow{d_{k}}$
\end_inset
are the positions of each point, with
\begin_inset Formula $\overrightarrow{d_{0}}=0$
\end_inset
.
So the intensity will be:
\begin_inset Formula \[
J(\overrightarrow{s})=\left|A(\overrightarrow{s})\right|^{2}=\sum_{j=0}^{N-1}\sum_{k=0}^{N-1}e^{2\pi i\overrightarrow{s}.(\overrightarrow{d_{j}}-\overrightarrow{d_{k}})}=N+1+2\sum_{j=1}^{N-1}\sum_{k
|
\begin_inset Text
\layout Standard
\begin_inset Graphics
filename figures/dirac_comb/dirac_comb.pgm
scale 60
\end_inset
\end_inset
|
|
\begin_inset Text
\layout Standard
\begin_inset Graphics
filename figures/dirac_comb/dirac_comb_fourier.pgm
scale 60
\end_inset
\end_inset
|
\end_inset
\layout Caption
\begin_inset LatexCommand \label{cap:Dirac-Comb}
\end_inset
Dirac Comb
\end_inset
\layout Subsubsection
The Diffraction Condition
\layout Standard
A Delone set is a crystal if its diffraction pattern has a countably infinite
discrete component.
\layout Subsection
The One Dimensional Case
\layout Standard
An infinite ordered set
\begin_inset Formula $\{ x_{k}\}=\dots,x_{-1},x_{0},x_{1},\dots$
\end_inset
,
\begin_inset Formula $x_{k}
|
\begin_inset Text
\layout Standard
\begin_inset Graphics
filename figures/sequence1/random.pgm
scale 75
\end_inset
\end_inset
|
|
\begin_inset Text
\layout Standard
\begin_inset Graphics
filename figures/sequence1/random_fourier.pgm
scale 75
\end_inset
\end_inset
|
|
\begin_inset Text
\layout Standard
random
\end_inset
|
\end_inset
\layout Standard
\begin_inset Tabular
|
\begin_inset Text
\layout Standard
\begin_inset Graphics
filename figures/sequence1/nonrandom.pgm
scale 75
keepAspectRatio
\end_inset
\end_inset
|
|
\begin_inset Text
\layout Standard
\begin_inset Graphics
filename figures/sequence1/nonrandom_fourier.pgm
scale 75
keepAspectRatio
\end_inset
\end_inset
|
|
\begin_inset Text
\layout Standard
periodic
\end_inset
|
\end_inset
\layout Standard
\begin_inset Tabular
|
\begin_inset Text
\layout Standard
\begin_inset Graphics
filename figures/sequence1/fibonacci.pgm
scale 75
keepAspectRatio
\end_inset
\end_inset
|
|
\begin_inset Text
\layout Standard
\begin_inset Graphics
filename figures/sequence1/fibonacci_fourier.pgm
scale 75
keepAspectRatio
\end_inset
\end_inset
|
|
\begin_inset Text
\layout Standard
Fibonacci
\end_inset
|
\end_inset
\layout Standard
\begin_inset Tabular
|
\begin_inset Text
\layout Standard
\begin_inset Graphics
filename figures/sequence1/paper.pgm
scale 75
keepAspectRatio
\end_inset
\end_inset
|
|
\begin_inset Text
\layout Standard
\begin_inset Graphics
filename figures/sequence1/paper_fourier.pgm
scale 75
keepAspectRatio
\end_inset
\end_inset
|
|
\begin_inset Text
\layout Standard
paper-folding
\end_inset
|
\end_inset
\layout Caption
\begin_inset LatexCommand \label{cap:method1transforms}
\end_inset
One Dimensional Lattices and their Fourier transforms
\end_inset
\layout Standard
All Fourier transforms have a bright point in the centre.
The random case has a more continuous diffraction pattern than the periodic
and Fibonacci cases.
The bright peaks for these two correspond to Bragg peaks.
Under this method the paper-folding lattice gives a very similar pattern
to the random case.
\layout Subsubsection
Method 2
\layout Standard
The letters in the sequence determine if there is a point or not on a regular
interval.
The results for the four sequences used are in figure
\begin_inset LatexCommand \ref{cap:method2transforms}
\end_inset
(see appendix
\begin_inset LatexCommand \ref{app:Commands-Used}
\end_inset
for commands used).
\layout Standard
\begin_inset Float figure
placement !h
wide false
collapsed true
\layout Standard
\begin_inset Tabular
|
\begin_inset Text
\layout Standard
\begin_inset Graphics
filename figures/sequence2/random.pgm
scale 75
\end_inset
\end_inset
|
|
\begin_inset Text
\layout Standard
\begin_inset Graphics
filename figures/sequence2/random_fourier.pgm
scale 75
\end_inset
\end_inset
|
|
\begin_inset Text
\layout Standard
random
\end_inset
|
\end_inset
\layout Standard
\begin_inset Tabular
|
\begin_inset Text
\layout Standard
\begin_inset Graphics
filename figures/sequence2/nonrandom.pgm
scale 75
keepAspectRatio
\end_inset
\end_inset
|
|
\begin_inset Text
\layout Standard
\begin_inset Graphics
filename figures/sequence2/nonrandom_fourier.pgm
scale 75
keepAspectRatio
\end_inset
\end_inset
|
|
\begin_inset Text
\layout Standard
periodic
\end_inset
|
\end_inset
\layout Standard
\begin_inset Tabular
|
\begin_inset Text
\layout Standard
\begin_inset Graphics
filename figures/sequence2/fibonacci.pgm
scale 75
keepAspectRatio
\end_inset
\end_inset
|
|
\begin_inset Text
\layout Standard
\begin_inset Graphics
filename figures/sequence2/fibonacci_fourier.pgm
scale 75
keepAspectRatio
\end_inset
\end_inset
|
|
\begin_inset Text
\layout Standard
Fibonacci
\end_inset
|
\end_inset
\layout Standard
\begin_inset Tabular
|
\begin_inset Text
\layout Standard
\begin_inset Graphics
filename figures/sequence2/paper.pgm
scale 75
keepAspectRatio
\end_inset
\end_inset
|
|
\begin_inset Text
\layout Standard
\begin_inset Graphics
filename figures/sequence2/paper_fourier.pgm
scale 75
keepAspectRatio
\end_inset
\end_inset
|
|
\begin_inset Text
\layout Standard
paper-folding
\end_inset
|
\end_inset
\layout Caption
\begin_inset LatexCommand \label{cap:method2transforms}
\end_inset
One Dimensional Lattices and their Fourier transforms
\end_inset
\layout Standard
Again the random case has a more continuous diffraction pattern than the
others.
The periodic almost has no continuous nature.
Both the Fibonacci and paper-folding patterns have discrete bright peaks.
\layout Section
The Tight-Binding Model
\layout Standard
\begin_inset LatexCommand \label{sec:The-Tight-Binding-Model}
\end_inset
This is a quantum model to describe a lattice of atoms, where we take the
case where the atoms are all considered separately, and consider what happens
if the wavefunctions overlap a little.
The time-independent Schr
\begin_inset ERT
status Collapsed
\layout Standard
\backslash
"o
\end_inset
dinger equation in Dirac notation is
\begin_inset Formula \[
\mathcal{H}\left|\Psi\right\rangle =E\left|\Psi\right\rangle \]
\end_inset
\layout Standard
where
\begin_inset Formula $\mathcal{H}$
\end_inset
is an
\emph on
operator
\emph default
acting on the
\emph on
eigenfunction
\emph default
\begin_inset Formula $\Psi$
\end_inset
which has
\emph on
eigenenergy
\emph default
\begin_inset Formula $E$
\end_inset
.
\begin_inset Formula $\Psi$
\end_inset
is a
\emph on
wavefunction
\emph default
which represents everything that we can know about the state of the system.
In the
\emph on
tight-binding method
\emph default
\begin_inset LatexCommand \cite{quantum,physics}
\end_inset
, we let
\begin_inset Formula $\mathcal{H}=\sum_{n}V_{n}a_{n}^{\dagger}a_{n}+\sum_{n}\sum_{m\neq n}t_{m,n}a_{m}^{\dagger}a_{n}$
\end_inset
, where
\begin_inset Formula $V_{n}$
\end_inset
is the potential at site
\begin_inset Formula $n$
\end_inset
, and
\begin_inset Formula $t_{m,n}$
\end_inset
corresponds the overlap mentioned above.
\begin_inset Formula $a_{n}^{\dagger}$
\end_inset
and
\begin_inset Formula $a_{n}$
\end_inset
are the creation and annihilation operators respectively for the site
\begin_inset Formula $n$
\end_inset
.
\begin_inset Formula $a_{n}^{\dagger}|0\rangle=|n\rangle$
\end_inset
and
\begin_inset Formula $\langle0|a_{n}=\langle n|$
\end_inset
.
\begin_inset Formula $|0\rangle$
\end_inset
represent the vacuum state, where all the atoms are in their ground state.
Thus we can represent the wavefunction as
\begin_inset Formula \[
|\Psi\rangle=\sum_{k}c_{k}a_{k}^{\dagger}|0\rangle\equiv\sum_{k}c_{k}|k\rangle\]
\end_inset
\layout Standard
and if we multiple the Schr
\begin_inset ERT
status Collapsed
\layout Standard
\backslash
"o
\end_inset
dinger equation on the left by
\begin_inset Formula $\langle0|a_{n}\equiv\langle n|$
\end_inset
we have, on the RHS,
\begin_inset Formula $\langle n|E\left|\Psi\right\rangle =E\sum_{k}c_{k}\langle n|k\rangle=E\sum_{k}c_{k}\delta_{n,k}=Ec_{n}$
\end_inset
.
For the LHS start with
\layout Standard
\begin_inset Tabular
|
\begin_inset Text
\layout Standard
\begin_inset Formula $\mathcal{H}|\Psi\rangle$
\end_inset
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $=\sum_{k}\sum_{l}c_{k}(V_{l}a_{l}^{\dagger}+\sum_{m\neq l}t_{l,m}a_{m}^{\dagger})a_{l}a_{k}^{\dagger}|0\rangle$
\end_inset
\end_inset
|
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $=\sum_{k}c_{k}V_{k}a_{k}^{\dagger}|0\rangle+\sum_{l}\sum_{m\neq k}c_{m}t_{l,m}a_{l}^{\dagger}|0\rangle$
\end_inset
\end_inset
|
\end_inset
\layout Standard
Now when we multiply on the left by
\begin_inset Formula $\langle n|\equiv\langle0|a_{n}$
\end_inset
we get
\layout Standard
\begin_inset Tabular
|
\begin_inset Text
\layout Standard
\begin_inset Formula $\langle n|\mathcal{H}|\Psi\rangle$
\end_inset
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $=\sum_{k}c_{k}V_{k}a_{k}^{\dagger}\langle0|a_{n}a_{k}^{\dagger}|0\rangle+\sum_{l}\sum_{m\neq l}c_{m}t_{l,m}\langle0|a_{n}a_{l}^{\dagger}|0\rangle$
\end_inset
\end_inset
|
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $=\sum_{k}c_{k}V_{k}\delta_{n,k}+\sum_{l}\sum_{m\neq l}c_{m}t_{l,m}\delta_{n,l}$
\end_inset
\end_inset
|
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $=V_{n}c_{n}+\sum_{m}t_{m,n}c_{m}$
\end_inset
\end_inset
|
\end_inset
\layout Standard
So we have (with
\begin_inset Formula $t_{m,n}=0$
\end_inset
for
\begin_inset Formula $n=m$
\end_inset
)
\begin_inset Formula \[
V_{n}c_{n}+\sum_{m}t_{m,n}c_{m}=Ec_{n}\]
\end_inset
\layout Standard
In the on-site diagonal model, we choose
\begin_inset Formula $t_{m,n}$
\end_inset
such that
\begin_inset Formula \[
V_{n}c_{n}+t(c_{n+1}+c_{n-1})=Ec_{n}\]
\end_inset
\layout Standard
that is, only the closest two atoms affect each atom.
We can write this in matrix form:
\begin_inset Formula $\left(\begin{array}{c}
c_{n+1}\\
c_{n}\end{array}\right)=T_{n}\left(\begin{array}{c}
c_{n}\\
c_{n-1}\end{array}\right)$
\end_inset
where
\begin_inset Formula $T_{n}=\left(\begin{array}{cc}
(E-V_{n})/t & -1\\
1 & 0\end{array}\right)$
\end_inset
.
Then, if we let
\begin_inset Formula $R_{n}=T_{n}T_{n-1}\cdots T_{2}T_{1}$
\end_inset
we have
\begin_inset Formula $c_{n}$
\end_inset
in terms of
\begin_inset Formula $c_{0}$
\end_inset
&
\begin_inset Formula $c_{1}$
\end_inset
.
Thus
\begin_inset Formula \[
\left(\begin{array}{c}
c_{n}\\
c_{n-1}\end{array}\right)=R_{n-1}\left(\begin{array}{c}
c_{1}\\
c_{0}\end{array}\right)\]
\end_inset
\layout Standard
Now, we consider the case when the potentials
\begin_inset Formula $\{ V_{n}\}$
\end_inset
are ordered in some aperiodic sequence.
That is, given a sequence
\begin_inset Formula $\{ a_{k}\}$
\end_inset
of some alphabet, be associate a potential with each letter of the alphabet.
for a binary alphabet, for example, we could have
\begin_inset Formula $0\longleftrightarrow+V$
\end_inset
,
\begin_inset Formula $1\longleftrightarrow-V$
\end_inset
.
\layout Subsection
Spectrum of the Model
\layout Standard
\begin_inset LatexCommand \label{sub:Spectrum}
\end_inset
The spectral theory associated to this model is different for the diffraction
patterns obtained in the aperiodic lattice model.
The spectrum for an operator
\begin_inset Formula $\mathcal{H}$
\end_inset
is defined as the complement of the set of energies for which the operator
\begin_inset Formula $(\mathcal{H}-E)^{-1}$
\end_inset
is bounded, that is
\begin_inset Formula \[
\sigma(\mathcal{H})=\overline{\left\{ E\in\mathbb{R}:(\mathcal{H}-E)^{-1}\textrm{is bounded}\right\} }\]
\end_inset
\layout Standard
If one thinks of the operators as matrices, the spectrum corresponds to
the eigenvalues
\begin_inset Formula $E$
\end_inset
for which
\begin_inset Formula $\mathcal{H}-EI$
\end_inset
has an inverse, that is its determinant is non-zero.
For the operator found above this is the same as the set of values of
\begin_inset Formula $E$
\end_inset
for which the operator
\begin_inset Formula $\mathcal{H}$
\end_inset
is polynomial bounded
\begin_inset LatexCommand \cite{quantum}
\end_inset
.
That is
\begin_inset Formula \[
\sigma(\mathcal{H})=\overline{\left\{ E\in\mathbb{R}:\mathcal{H}(c_{n})=Ec_{n}\textrm{and}\left|c_{n}\right|\leq C\left|n\right|^{a}\forall n\in\mathbb{Z}\right\} }\]
\end_inset
\layout Standard
This spectrum is composed of three parts: the pure point , the absolutely
continuous, and the singular continuous.
The pure point corresponds to the allowed energy eigenvalues of the equation,
the absolutely continuous part is a closed set with non-empty interior,
while the singular continuous is what is left.
\layout Subsection
Bloch Condition
\layout Standard
If the wave function were periodic, it would satisfy the Bloch condition
\begin_inset LatexCommand \cite{physics}
\end_inset
:
\layout Standard
\begin_inset Formula \[
\Psi_{n}(\mathbf{r})=\sum_{R}e^{ik.R}\Psi_{n}(\mathbf{r}-\mathbf{R})\]
\end_inset
\layout Standard
\begin_inset Tabular
|
\begin_inset Text
\layout Standard
\begin_inset Formula $\Psi(\mathbf{r}+\mathbf{R})$
\end_inset
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $=\sum_{R^{'}}e^{ik.R^{'}}\Psi_{n}(\mathbf{r}+\mathbf{R}^{'}-\mathbf{R})$
\end_inset
\end_inset
|
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $=e^{ik.R}\sum_{R^{'}}e^{ik.(R^{'}-R)}\Psi_{n}(\mathbf{r}-(\mathbf{R}^{'}-\mathbf{R}))$
\end_inset
\end_inset
|
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $=e^{ik.R}\sum_{R^{'}}e^{ik.R^{''}}\Psi_{n}(\mathbf{r}-\mathbf{R}^{''})$
\end_inset
\end_inset
|
|
\begin_inset Text
\layout Standard
\end_inset
|
\begin_inset Text
\layout Standard
\begin_inset Formula $=e^{ik.R}\Psi(\mathbf{r})$
\end_inset
\end_inset
|
\end_inset
\layout Standard
In the aperiodic case it would approximately satisfy the Bloch condition.
In terms of the co-efficients
\begin_inset Formula $c_{k}$
\end_inset
this condition is
\layout Standard
\begin_inset Formula \[
c_{j+n}=e^{ikn}c_{j}\]
\end_inset
\layout Standard
Here, the period of the wavefunction is
\begin_inset Formula $n$
\end_inset
, compare with the value
\begin_inset Formula $R$
\end_inset
above.
If the period was
\begin_inset Formula $1$
\end_inset
, i.e.
\begin_inset Formula $V_{n}=V\;\forall n$
\end_inset
then we have
\begin_inset Formula $c_{j+1}=e^{ik}c_{j}$
\end_inset
.
Substituting this into
\begin_inset Formula $V_{n}c_{n}+t(c_{n+1}+c_{n-1})=Ec_{n}$
\end_inset
then dividing through by
\begin_inset Formula $c_{n}$
\end_inset
gives us
\begin_inset Formula $V+t(e^{ik}+e^{-ik})=E$
\end_inset
, that is
\begin_inset Formula $(E-V)/t=2\cos k$
\end_inset
.
To the trace
\begin_inset Formula $\textrm{tr}T=2\cos k$
\end_inset
.
In which case the energy spectrum is
\begin_inset Formula $\sigma(\mathcal{H})=\{ E\in\mathbb{R}\,:\,|\textrm{tr}T|\leq2\}$
\end_inset
Similarly, for the case with period
\begin_inset Formula $n$
\end_inset
we have
\begin_inset Formula $\textrm{tr}R_{n}=2\cos kn$
\end_inset
, so the energy spectrum is
\begin_inset Formula $\sigma(\mathcal{H})=\{ E\in\mathbb{R}\,:\,|\textrm{tr}R_{n}|\leq2\}$
\end_inset
.
\layout Standard
For the aperiodic case, we consider a section of the sequence of length
\begin_inset Formula $n$
\end_inset
as the basis for a periodic sequence, then let
\begin_inset Formula $n\rightarrow\infty$
\end_inset
.
This spectrum is called the
\emph on
dynamical
\emph default
spectrum.
\layout Subsection
Fibonacci Sequence
\layout Standard
To calculate the energy spectrum of the Fibonacci sequence, note that due
to the way the sequence is constructed,
\begin_inset Formula $R_{n}=R_{n-2}R_{n-1}$
\end_inset
.
From this equation, we find a trace map for the Fibonacci model:
\begin_inset Formula \[
\textrm{tr}R_{n}=\textrm{tr}R_{n-2}.\textrm{tr}R_{n-1}-\textrm{tr}R_{n-3}\]
\end_inset
\layout Standard
This map can then be used to find dynamical energy spectrum.
This is beyond the scope of this project.
\layout Standard
Then, given the energy spectrum, that is the allowed energies, we can then
find the corresponding wavefunctions.
This can be done using the recurrence relation
\begin_inset Formula \[
c_{n+1}=c_{n}(E-V_{n})/t-c_{n-1}\]
\end_inset
\layout Standard
As we can take the allowed values for
\begin_inset Formula $E$
\end_inset
from the dynamical spectrum, and the values of
\begin_inset Formula $V_{n}$
\end_inset
are given by our chosen aperiodic sequence.
The wavefunction is then given by
\begin_inset Formula \[
|\Psi\rangle=\sum_{k}c_{k}a_{k}^{\dagger}|0\rangle\equiv\sum_{k}c_{k}|k\rangle\]
\end_inset
\layout Subsection
The Wavefunctions Obtained
\layout Standard
We can calculate the wavefunctions for each energy.
Not all energies are allowed; the spectrum, as described in section
\begin_inset LatexCommand \ref{sub:Spectrum}
\end_inset
, tells us which energies will give us a bounded wavefunction.
If a disallowed energy value is used, the wavefunction will just tend to
infinity.
For periodic case I found periodic wavefunction like the one show in figure
\begin_inset LatexCommand \ref{cap:Wavefunction-for-Periodic}
\end_inset
.
I was unable to find suitable energies for the aperiodic cases however.
This is because the allowed energies are not continuous, but a
\emph on
cantor dust
\emph default
.
Thus the accuracy of the energy value used becomes very import and the
slightest difference will not be an allowed energy.
Figure
\begin_inset LatexCommand \ref{cap:Wavefunction-for-Paper}
\end_inset
shows, for a disallowed energy, how the wavefunction grows unbounded.
\layout Standard
\begin_inset Float figure
placement H
wide false
collapsed true
\layout Standard
\begin_inset Graphics
filename c/spectrums/spectrum3.ps
lyxscale 50
scale 40
rotateAngle 270
\end_inset
\layout Caption
\begin_inset LatexCommand \label{cap:Wavefunction-for-Periodic}
\end_inset
Wavefunction for Periodic potential
\end_inset
\begin_inset Float figure
placement H
wide false
collapsed true
\layout Standard
\begin_inset Graphics
filename c/spectrums/spectrum7.ps
lyxscale 50
scale 40
rotateAngle 270
\end_inset
\layout Caption
\begin_inset LatexCommand \label{cap:Wavefunction-for-Paper}
\end_inset
Wavefunction for paperfolding potential
\end_inset
\layout Section
Conclusion
\layout Standard
We have seen that the paperfolding sequence has several interesting properties,
especially when expressed as a dragon curve.
We have also seen how it can be used with continued fractions and that
it could have applications in the study of quasicrystals.
Unfortunately I was not able to graph the wavefunction in the case of the
[Fibonacci or] paperfolding sequence.
\layout Subsection
Acknowledgments
\layout Standard
I would like to thank Dr Chris J Smyth for his invaluable help in my undertaking
of this project.
\layout Section
\start_of_appendix
Source Code
\layout Standard
Attached is the source code for the programs that I wrote to generate the
diagrams used in this project.
fourier.c and points2image.c are based on a program fourier.c in
\begin_inset LatexCommand \cite{quasicrystal}
\end_inset
.
\layout Paragraph
C programs:
\layout List
\labelwidthstring 00.00.0000
project.h A header file common to most of the c programs.
Contains some constants and macros.
\layout List
\labelwidthstring 00.00.0000
dragonplot.c Takes a binary series as instruction to turn left/right when
drawing a dragon curve.
See section
\begin_inset LatexCommand \ref{cap:Fractal}
\end_inset
.
Outputs a .pgm image.
\layout List
\labelwidthstring 00.00.0000
dragonplot2.c As above, but outputs an xfig .fig file.
\layout List
\labelwidthstring 00.00.0000
fibbonacci.c Generates the Fibonacci sequence.
Uses the method described for the Sturmien sequence.
\layout List
\labelwidthstring 00.00.0000
fold.c Outputs a paperfolding sequence based on the folding instructions
given.
\layout List
\labelwidthstring 00.00.0000
fourier.c Ouputs an image of the fourier transform of a given set of points.
\layout List
\labelwidthstring 00.00.0000
image2dat1d.c Creates a graph of the horizontal cross-section of a given
image.
\layout List
\labelwidthstring 00.00.0000
image2fourier.c Calculates the fourier transform of the input image.
Uses
\emph on
fftw
\emph default
: the
\begin_inset Quotes eld
\end_inset
Fastest Fourier Transform in the West
\begin_inset Quotes erd
\end_inset
(sic)
\layout List
\labelwidthstring 00.00.0000
newton.c Calculates the solution to
\begin_inset Formula $\lambda^{3}-\lambda^{2}-2=0$
\end_inset
using Newton's method.
\layout List
\labelwidthstring 00.00.0000
nonrandom.c Generates the sequence
\begin_inset Formula $ababababa\dots$
\end_inset
\layout List
\labelwidthstring 00.00.0000
points2image.c Draws a picture of points in space given by the input.
\layout List
\labelwidthstring 00.00.0000
random.c Generates a random binary sequence.
\layout List
\labelwidthstring 00.00.0000
sequence2points.c Calculates positions of points based on method 1, see section
\begin_inset LatexCommand \ref{sub:make1dlattice}
\end_inset
.
\layout List
\labelwidthstring 00.00.0000
sequence2points2.c Calculates positions of points based on method 2, see
section
\begin_inset LatexCommand \ref{sub:make1dlattice}
\end_inset
.
\layout List
\labelwidthstring 00.00.0000
sequence2spectrum.c Graphs the wavefunction with the potential
\begin_inset Formula $V_{n}$
\end_inset
determined by the sequence given, see section
\begin_inset LatexCommand \ref{sec:The-Tight-Binding-Model}
\end_inset
.
\layout Paragraph
Logo Programs:
\layout List
\labelwidthstring 00.00.0000
heighway.lg Used to draw the dragon curve's boundary.
Can also draw the dragon curve itself.
\layout Subsection
How they work
\layout Standard
I made the programs very modular to that I could use pipes to create a variety
of different configurations to try out.
Most of the programs work by taking information in from the standard input
(
\emph on
stdin
\emph default
), processing it, and then outputting different data to the standard out
(
\emph on
stdout
\emph default
).
\emph on
Pipes
\emph default
are used to link together these programs in a `chain'.
\layout Standard
For example, if I want to see what the dragon curve looks like, I use
\emph on
echo
\emph default
to create the folding instructions
\begin_inset Formula $aaaaaaaaaaaa$
\end_inset
, then pipe this into the folding program
\emph on
fold
\emph default
.
It outputs the paper-folding sequence, which I pipe into
\emph on
dragonplot
\emph default
to draw the dragon curve, and display the result using
\emph on
xv
\emph default
.
The command for this would be
\layout Quote
echo aaaaaaaaaaaaa | ./fold | ./dragonplot 400 500 2 100 150 | xv -
\layout Standard
Which would generate the image show in figure
\begin_inset LatexCommand \ref{cap:./dragonplot}
\end_inset
.
The options dragonplot is called with set the size of the output image
etc.
\layout Standard
\begin_inset Float figure
placement h
wide false
collapsed true
\layout Standard
\begin_inset Graphics
filename figures/dragonplot.png
lyxscale 50
scale 40
\end_inset
\layout Caption
\begin_inset LatexCommand \label{cap:./dragonplot}
\end_inset
./dragonplot
\end_inset
\layout Subsubsection
Commands Used
\layout Standard
\begin_inset LatexCommand \label{app:Commands-Used}
\end_inset
The commands used to generate the images in figure
\begin_inset LatexCommand \ref{cap:method1transforms}
\end_inset
are
\layout Quotation
random 65 |sequence2points 25 |points2image 480 50 |xv -
\layout Quotation
random 65 |sequence2points 25 |fourier 480 50 96 2 |xv -
\layout Quotation
nonrandom 65 |sequence2points 25 |points2image 480 50 |xv -
\layout Quotation
nonrandom 65 |sequence2points 25 |fourier 480 50 96 2 |xv -
\layout Quotation
fibbonacci 58 |sequence2points 25 |points2image 480 50 |xv -
\layout Quotation
fibbonacci 58 |sequence2points 25 |fourier 480 50 96 2 |xv -
\layout Quotation
paper 6 |sequence2points 25 |points2image 480 50 |xv -
\layout Quotation
paper 6 |sequence2points 25 |fourier 480 50 96 2 |xv -
\layout Standard
and for figure
\begin_inset LatexCommand \ref{cap:method2transforms}
\end_inset
\layout Quotation
random 100 |sequence2points2 25 |points2image 480 50 |xv -
\layout Quotation
random 100 |sequence2points2 25 |fourier 480 50 96 2 |xv -
\layout Quotation
nonrandom 100 |sequence2points2 25 |points2image 480 50 |xv -
\layout Quotation
nonrandom 100 |sequence2points2 25 |fourier 480 50 96 2 |xv -
\layout Quotation
fibbonacci 100 |sequence2points2 25 |points2image 480 50 |xv -
\layout Quotation
fibbonacci 100 |sequence2points2 25 |fourier 480 50 96 2 |xv -
\layout Quotation
paper 7 |sequence2points2 25 |points2image 480 50 |xv -
\layout Quotation
paper 7 |sequence2points2 25 |fourier 480 50 96 2 |xv -
\layout Subsubsection
fold.c
\layout Standard
This program implements the 2-automaton described in section
\begin_inset LatexCommand \ref{sub:2-Automaton}
\end_inset
, but with the modification for other folding instructions, as described
in section
\begin_inset LatexCommand \ref{peanoinstr}
\end_inset
.
\layout Subsubsection
fibonacci.c
\layout Standard
The Fibonacci sequence is a Sturmian sequence.
This program uses the method described in section
\begin_inset LatexCommand \ref{sub:Sturmian}
\end_inset
, with
\begin_inset Formula $\alpha=\phi-1=\frac{\sqrt{5}-1}{2}$
\end_inset
and
\begin_inset Formula $\beta=0$
\end_inset
as per section
\begin_inset LatexCommand \ref{sub:Fibonnaci}
\end_inset
.
\layout Subsubsection
newton.c
\layout Standard
This solves the equation
\begin_inset Formula $g(\lambda)=\lambda^{3}-\lambda^{2}-2=0$
\end_inset
.
Let
\begin_inset Formula $g(x)/g^{'}(x)=\frac{x^{3}-x^{2}-2}{3x^{2}-3x}$
\end_inset
.
Then define the recurrence relation
\begin_inset Formula $x_{k+1}=x_{k}-g(x_{k})/g^{'}(x_{k})$
\end_inset
.
With a suitible choice for
\begin_inset Formula $x_{0}$
\end_inset
,
\begin_inset Formula $\lim_{k\rightarrow\infty}x_{k}=\lambda$
\end_inset
where
\begin_inset Formula $\lambda$
\end_inset
is a solution to
\begin_inset Formula $g(\lambda)=0$
\end_inset
.
That is what newton.c does.
Running ./newton has the following output:
\layout Quotation
x0 = 3.000000
\layout Quotation
x1 = 2.238095
\layout Quotation
x2 = 1.839868
\layout Quotation
x3 = 1.709679
\layout Quotation
x4 = 1.695773
\layout Quotation
x5 = 1.695621
\layout Quotation
x6 = 1.695621
\layout Quotation
x7 = 1.695621
\layout Quotation
x8 = 1.695621
\layout Quotation
x9 = 1.695621
\layout Standard
So the solution we are looking for is 1.69562 to six decinal places.
\layout Subsection
Logo
\layout Standard
The logo programming language is well suited to drawing the dragon curves
in this project.
The language works by imagining that there is a turtle, with a pen taped
to it, on a piece of paper.
The language is made up of instruction like
\begin_inset Quotes eld
\end_inset
forward 50
\begin_inset Quotes erd
\end_inset
and
\begin_inset Quotes eld
\end_inset
left 45
\begin_inset Quotes erd
\end_inset
.
This tells the `turtle' to move, and so draw out a line on the paper.
The logo interpreter xlogo is shown in figure
\begin_inset LatexCommand \ref{cap:xlogo.jar}
\end_inset
.
\layout Standard
\begin_inset Float figure
placement h
wide false
collapsed true
\layout Standard
\begin_inset Graphics
filename figures/xlogo.png
lyxscale 50
scale 40
\end_inset
\layout Caption
\begin_inset LatexCommand \label{cap:xlogo.jar}
\end_inset
xlogo.jar
\end_inset
\layout Standard
They way my logo program
\emph on
heighway.lg
\emph default
draws the graphs is very simple.
It is basically the Maudlin-Williams graph with information about
\emph on
direction
\emph default
as well as scale.
Using this, and only this, the turtle will draw out the desired fractals.
For example, the boundary is made up of two segments,
\emph on
a
\emph default
and
\emph on
b
\emph default
(see section
\begin_inset LatexCommand \ref{sub:Boundary-of-Dragon}
\end_inset
).
\emph on
a
\emph default
itself is made up of a copy of
\emph on
a
\emph default
and of
\emph on
b
\emph default
,
\begin_inset Formula $\frac{1}{\sqrt{2}}$
\end_inset
the length of the original
\emph on
a
\emph default
, but at a
\begin_inset Formula $45^{o}$
\end_inset
degree angle to the original
\emph on
a.
b
\emph default
is made up of two half-size
\emph on
a
\emph default
s, but with one pointing in the opposite direction.
\layout Section
Software
\layout List
\labelwidthstring 00.00.0000
Gentoo This project was done on my PC running Gentoo GNU/Linux (http://www.gentoo.
org)
\layout List
\labelwidthstring 00.00.0000
LyX This project was written using the LyX word processor (http://www.lyx.org/)
\layout List
\labelwidthstring 00.00.0000
xfig Figures were drawn using xfig (http://www.xfig.org)
\layout List
\labelwidthstring 00.00.0000
gimp Some diagrams were drawn/edited using the gimp (http://www.gimp.org/)
\layout List
\labelwidthstring 00.00.0000
imagemagick Batch image manipulation (http://www.imagemagick.org/)
\layout List
\labelwidthstring 00.00.0000
xv Image viewing and manipulation (http://www.trilon.com/xv/)
\layout List
\labelwidthstring 00.00.0000
a2ps Used to format source code for printing
\layout List
\labelwidthstring 00.00.0000
logo Berkeley Logo interpreter (UCBLogo) (http://www.cs.berkeley.edu/~bh/)
\layout List
\labelwidthstring 00.00.0000
xlogo Colourful java logo interpreter (http://xlogo.free.fr/)
\layout List
\labelwidthstring 00.00.0000
fftw The
\begin_inset Quotes eld
\end_inset
Fastest Fourier Transform in the West
\begin_inset Quotes erd
\end_inset
(http://www.fftw.org)
\layout Bibliography
\bibitem [1]{folds}
\emph on
Macquarie Mathematics Report -
\begin_inset Quotes eld
\end_inset
Folds!
\begin_inset Quotes erd
\end_inset
\emph default
- Michel Dekking, Michel Mendes France, Alf van der Poorten (1982)
\layout Bibliography
\bibitem [2]{maxpaper}
\emph on
How To Fold Paper in Half Twelve Times
\emph default
- Britney Gallivan ( http://www.osb.net/Pomona/12times.htm)
\layout Bibliography
\bibitem [3]{fibonaccipdf}
\emph on
Fibonacci Sequences Arising From Paperfolding
\emph default
- Anna Rodenhausen (http://www.mevis.de/~anna/publications/goslar.pdf )
\layout Bibliography
\bibitem [4]{fractions}
\emph on
Folded Continued Fractions
\emph default
- A.
J.
van der Poorten and J.
Shallit (http://www-centre.mpce.mq.edu.au/alfpapers/a098.pdf)
\layout Bibliography
\bibitem [5]{automata}
\emph on
Finite Automata and Arithmetic
\emph default
- J.-P.
Allouche (http://www.emis.de/journals/SLC/opapers/s30allouche.pdf)
\layout Bibliography
\bibitem [6]{quantum}
\emph on
Aperiodically Ordered Structures in One Dimension
\emph default
- Michael Hornquist (http://www.ifm.liu.se/~micho/phd/dissertation.html)
\layout Bibliography
\bibitem [7]{jurassic}
\emph on
Jurassic Park
\emph default
- Michael Crichton
\layout Bibliography
\bibitem [8]{Fractal Geometry}
\emph on
Fractal Geometry
\emph default
- Kenneth Falconer
\layout Bibliography
\bibitem [9]{Measure}
\emph on
Measure, Topology, and Fractal Geometry
\emph default
- Gerald A.
Edgar
\layout Bibliography
\bibitem [10]{quasicrystal}
\emph on
Quasicrystals and Geometry
\emph default
- Marjorie Senechal
\layout Bibliography
\bibitem [11]{aperiodic}
\emph on
Almost Periodic Sequnces
\emph default
- Harald Bohr
\layout Bibliography
\bibitem [12]{physics}
\emph on
Solid State Physics
\emph default
- Neil W.
Ashcroft & N.
David Mermin
\layout Bibliography
\bibitem [13]{mathworld}
http://www.mathworld.com
\layout Bibliography
\bibitem [14]{wikipedia}
http://www.wikipedia.org
\layout Bibliography
\bibitem [15]{scienceworld}
http://www.scienceworld.com
\layout Bibliography
\bibitem [16]{fourier}
\emph on
School of Physics: The Fourier Transform (What you need to know)
\emph default
- (http://www.ph.ed.ac.uk/~wjh/teaching/Fourier/)
\the_end