100% found this document useful (1 vote)

134 views

Classical Math Fractals in PostScript

Classical mathematical fractals in BASIC are explained and converted into mean-and-lean EPSF def s, of which the .eps pictures are delivered in .pdf format and cropped to the prescribed BoundingBox when processed by Acrobat Pro , to be included easily in pdf(La)TEX, Word, ... documents. The EPSF fractals are transcriptions of the Turtle Graphics BASIC codes or programmed anew, recursively, based on the production rules of oriented objects. The Linden- mayer production rules are enriched by Pos

Uploaded by

jay

Available Formats

Download as PDF, TXT or read online on Scribd

100% found this document useful (1 vote)

134 views

Classical Math Fractals in PostScript

Uploaded by

jay

Available Formats

Download as PDF, TXT or read online on Scribd

You are on page 1/ 30

Kees van der Laan VOORJAAR 2013 49

Classical Math Fractals in PostScript

Fractal Geometry I

Abstract
Classical mathematical fractals in BASIC are explained and converted into mean-and-lean EPSF
defs, of which the .eps pictures are delivered in .pdf format and cropped to the prescribed
BoundingBox when processed by Acrobat Pro, to be included easily in pdf(La)TEX, Word, …
documents. The EPSF fractals are transcriptions of the Turtle Graphics BASIC codes or pro-
grammed anew, recursively, based on the production rules of oriented objects. The Linden-
mayer production rules are enriched by PostScript concepts. Experience gained in converting
a TEX script into WYSIWYG Word is communicated.

Keywords
Acrobat Pro, Adobe, art, attractor, backtracking, BASIC, Cantor Dust, C curve, dragon curve,
EPSF, FIFO, fractal, fractal dimension, fractal geometry, Game of Life, Hilbert curve, IDE (In-
tegrated development Environment), IFS (Iterated Function System), inﬁnity, kronkel (twist),
Lauwerier, Lévy, LIFO, Lindenmayer, minimal encapsulated PostScript, minimal plain TeX,
Minkowski, Monte Carlo, Photoshop, production rule, PSlib, self-similarity, Sierpiński (island,
carpet), Star fractals, TACP, TEXworks, Turtle Graphics, (adaptable) user space, von Koch
(island), Word

Contents
󰀭 Introduction
󰀭 Lévy (Properties, PostScript program, Run the program, Turtle Graphics) Cantor0
󰀭 Lindenmayer enriched by PostScript concepts for the Lévy fractal
󰀭 von Koch (Properties, PostScript def, Turtle Graphics, von Koch island) Cantor1
󰀭 Lindenmayer enriched by PostScript concepts for the von Koch fractal Cantor2
󰀭 Kronkel Cantor3
󰀭 Minkowski
󰀭 Dragon ﬁgures
󰀭 Stars
󰀭 Game of Life
󰀭 Annotated References
󰀭 Conclusions (TEX mark up, Conversion into Word)
󰀭 Acknowledgements (IDE)
󰀭 Appendix: Fractal Dimension
󰀭 Appendix: Cantor Dust
󰀭 Appendix: Hilbert Curve
Peano curves: order 1, 2, 3
󰀭 Appendix: Sierpiński islands

Introduction
My late professor Hans Lauwerier published nice, inspiring booklets about fractals
with programs in BASIC. However, I don’t know how to include elegantly the pic-
tures, obtained by running the BASIC codes, in my documents. Moreover, I consider
PostScript (PS, for short) more portable in place and time, can include EPSF results in
my TEX documents1 easily, and … do realize that PS is the de-facto standard industrial
printer language.
50 MAPS 44 Kees van der Laan

This note is about conversion of some of Lauwerier’s BASIC Turtle Graphics codes
for the simplest fractals into EPSF, ànd about the programming of new recursive EPSF
defs biased by Lindenmayer production rules for oriented objects, enriched with PS
concepts.

← Hilbert curves

Sierpiński →
islands 1, 2, 3

Now and then I have explained Lauwerier’s algorithms, especially when he asso-
ciates binary and quaternary number representations with fractals.
Fractals have widened the dimension concept into fractal-valued dimensions. Al-
though the fractal dimension concept is not necessary in order to understand the
codes, I have added the appendix Fractal Dimension, because fractal dimensions con-
tribute to characterizing fractals. Moreover, fractal dimension gives meaning to the
19th century ‘monstruous’ plane-filling curves.
Fractals were invented in the 20th century, and became the geometry of this cen-
tury due to the development of computers, because computers are the tools for view-
ing and researching fractals.
The ancestor of fractals is the 1D Cantor Dust. 2D predecessors of fractals are
the plane-filling curves named after Peano, Hilbert, Sierpiński, … , which captivated
mathematicians in the late 19th and the early 20th century.
Sierpiński curves have found their niche in the solution of the travelling salesman
problem.
In the sequel Lévy, von Koch, Kronkel (Dutch, means twist), Minkowski, Dragon
curve, star fractals, and a variant of the Game of Life are discussed. There are 4 ap-
pendices: the first about Fractal Dimension, the second about the historical Cantor
Dust, the third about the classical Hilbert curve, and the last about Sierpiński islands.
In the footsteps of Lauwerier, the reader is invited to experiment with the PS
programs, of which defs are supplied in my PSlib.eps library, which I’ll send on
request. MetaPost aficionados may translate the included Metafont codes into Meta-
Post, I presume.

Lévy fractal
An approximation of the Lévy fractal is also called a C (broken) line of a certain
order. The constructive definition of various orders of C lines starts with a straight
line, let us call this line C0 . An isosceles triangle with angles 45∘ , 90∘ and 45∘ is
built on this line as hypotenuse. The original line is then replaced by the other two
sides of this triangle to obtain C1 . Next, the two new lines each form the base for
another right-angled isosceles triangle, and are replaced by the other two sides of
their respective triangle, to obtain C2 . After two steps, the broken line has taken the
appearance of three sides of a rectangle of twice the length of the original line. At
each subsequent stage, each segment in the C figure is replaced by the other two
sides of a right-angled isosceles triangle built on it. Such a rewriting relates to a
Lindenmayer system. After n stages the C line has length 2n/2 × C0 : 2n segments
each of size 2−n/2 × C0 .
Fractals have various infinite lengths. The question arose: Can these blends of ∞
be used to characterize fractals?2 Below C0 … C6 and C10 have been constructed
from the definition.
Classical Math Fractals in PostScript VOORJAAR 2013 51

order 0 order 1 order 2 order 3 order 4 order 5 order 6 order 10

Properties
1a. The above sequence of curves loosely obey
Ci = C45 −45
i−1 ⊕ Ci−1 , i = 1, 2, … C0 = segment
⊕ means spliced i−1 means rotated over 45 .
C45 ∘

In fractal terminology such a recursion, or production rule, characterizes what

is called the self-similarity of fractals, because Ci consists of 2 spliced copies
of Ci−1 , which are not scalars but 2D, oriented objects. (Positive rotation is
counter-clockwise à la PS). Lindenmayer (Dutch theoretical biologist) invented
production rules in order to describe plants; production rules are also used in
program development.

1b. The C curves at right have a diﬀerent orientation.

The formula, which reﬂects the self-similarity in this orientation,
reads
Ci = Ci−1 ⊕ C−90
i−1 , i = 1, 2, …
⊕ means spliced i−1 means rotated over − 90 .
C−90 ∘

Self-similarity as construction method can be suitably programmed in Meta-

Post/-font, with their path data structure, as follows. Create the row of paths
p0 , p1 , p2 , …

p0 = C 0 , p45 −45 → p
i ⊕pi i+1 , for i = 0, 1, 2, … with ⊕ the splice operator.

2. In the pen-plotter days the natural question arose: What is the direction of a
segment? Lauwerier(1987) gives the intriguing relationship between the an-
gle ϕk of a segment and its index k (according to the orientation as given un-
der 1b).
π
ϕk = (sk mod 4) with sk = ∑p−1 b
j=0 j
sum of bimals of k
2
p−1
and k = ∑j=0 bj 2
j binary representation of k.

3. The Lévy fractal has fractal dimension 2, a local plane-ﬁlling curve, Lauw-
erier(1990). The C curves intersects themselves from order 4 onward.
The PostScript program
One might create an eﬃcient recursive backtracking program based on property 1a,
as a levyC def with the def given below. Scaling is commented out; just remove the
two % signs if scaling is wanted.

%!PS-Adobe-3.0 EPSF-3.0
%%Author: Kees van der Laan
%%Date: april 2011
%%Affiliation: kisa1@xs4all.nl
%%BoundingBox: -1 -1 346 61
%%BeginSetup %crops to BoundingBox
%%EndSetup %by Acrobat Pro
%%BeginProlog%collection of defs
/levyC{%on stack: the order ==> C line
52 MAPS 44 Kees van der Laan

%s = size of line segment (global)

dup 0 eq
{0 0 moveto s 0 lineto currentpoint stroke translate}%draw line
{1 sub %/s s 1.4142 div def %lower order on stack; s scaled variant
45 rotate levyC%-45 rotate%combine -45 twice into -90
-90 rotate levyC 45 rotate
1 add %/s s 1.4142 div def %adjust order on stack, and s(cale)
}ifelse
}def
%%EndProlog
%
% Program --- the script ---
%
/s 20 def 0 levyC pop
s 2 div 0 translate 1 levyC pop
s 0 translate 2 levyC pop
1.5 s mul 0 translate 3 levyC pop
2.5 s mul 0 translate 4 levyC pop
showpage
%%EOF

0 1 order 2 order 3 order 4

The above mean-and-lean PS def is the result of programming in the spirit of The
Art of Computer Programming, TACP for short. I’ll come back on a more systematic
approach of programming based on production rules, a little further on.
If the levyC def is included in the PSlib.eps library, then the above def can be
replaced by
(C:\\PSlib\\PSlib.eps) run
This feature of the run command is not generally known, so it seems.
Because programming in PostScript is subtle, I have included below the PS def
based on the production rule as stated under property 1b, which is highly similar to
the above backtracking process, but the result diﬀers in orientation. This levyCvar
def is also included in PSlib.eps.
/levyCvar{%order on stack ==> C line
%s = size of segment (global))
dup 0 eq
{0 0 moveto s 0 lineto
currentpoint stroke translate}%draw line
{1 sub %lower order on stack
levyCvar %C line
-90 rotate levyCvar 90 rotate %rotated C line
1 add %adjust order on stack
}ifelse }def

To run the program store the file with extension .eps (or .ps), right-mouse click
the thumbnail of the file and choose the option convert to Adobe PDF in the pop-up
menu. That is all when you have installed Acrobat Pro 7. (Other versions of Creative
Suite ask for open in Acrobat.) I also used Adobe Illustrator and PSView. The latter
just by double-clicking the filename upon which the command window opened and
a little later PSview.3
The Turtle Graphics algorithm is based on property 2. In order to understand the
formula mentioned, a table for the direction (with orientation as mentioned under
property 1b) of each segment is included. Such a table forms the basis for discovering
the regularity.
Classical Math Fractals in PostScript VOORJAAR 2013 53

k
order 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
0 →
1 → ↓
2 → ↓ ↓ ←
3 → ↓ ↓ ← ↓ ← ← ↑
4 → ↓ ↓ ← ↓ ← ← ↑ ↓ ← ← ↑ ← ↑ ↑ →

sk mod 4 0 1 1 2 1 2 2 3 1 2 2 3 2 3 3 0 …

Below I have included Lauwerier’s program and my conversion in PS, which is in-
teresting because of the transformation of the user space by ϕk , k = 0, 1, 2, … .

10 REM Levy fractal %!PS-Adobe-3.0 EPSF-3.0

10 P=12 : REM***order*** %%Title: Levy fractal a la Lauwerier
20 H=2^(-(P/2)) : A=H*COS(P*PI/4) : B=H*SIN(P*PI/4) %%Transcriptor: Kees van der Laan, April 2011, kisa1@xs4all.nl
30 LINE (0,0)-(A,-B) : LINE -(A+B,A-B) /LevyLauwerier{/p exch def%order =>C curve (l is global)
40 X=1 : Y=1 0 0 moveto l 0 lineto %order 0
50 FOR N=2 TO 2^P-1 p 1 eq {0 l neg rlineto} %order 1, segments 0, 1
60 M=N : S=1 {0 %previous phi on stack
70 IF M MOD 2 = 1 THEN S=S=1 1 1 2 p exp 1 sub {%for%n, the number of the segments 1,2,...
80 M\2 0 exch %s n :s sum of bimals and n the segment number on stack
90 IF M>1 THEN GOTO 70 p{dup 2 mod 3 -1 roll add exch 2 idiv}repeat pop%discard n
100 IF S MOD 4 = 0 THEN X=X+1 4 mod 90 mul dup 3 1 roll sub rotate l 0 rlineto
110 IF S MOD 4 = 1 THEN Y=Y+1 }for%n
120 IF S MOD 4 = 2 THEN X=X-1 pop%discard phi
130 IF S MOD 4 = 3 THEN Y=Y-1 }ifelse stroke}def
140 LINE -(A*X+B*Y, A*Y-B*X) /l 5 def 10 LevyLauwerier stroke showpage
150 NEXT N %%EOF
160 END

Size of line-piece is wired-in (In PS parametrized)

Subtle stack programming with only 1 rotate for each segment instead of a rotate
and a back rotate. The length l of each segment is a global parameter.
Below at left C0 …C5 , and C10 ; at right C10 spliced with its -1 1 scaled copy, a
Lévy carpet.

In my PWT guide of 1995, I did program the above Lévy fractal in TEX (orientation
1b) by the Turtle graphics method, in the footsteps of Knuth. Nowadays, I much
prefer the much more powerful and useful PostScript for programming my graphics.
Sorry to say so, but Knuth put me on the wrong track by his graphics in the TEXbook.
Lindenmayer enriched by PostScript concepts for the Lévy fractal
What we miss in the 1a property speciﬁcation is the scaling to smaller size of the
segments when the order increases, as well as a more precise meaning of what spliced
entails. A more accurate and improved production rule à la 1a, can be obtained when
we use PS concepts in the production rule at the expense of simplicity.
54 MAPS 44 Kees van der Laan

Cn = [R45 S( 1
, 1 ) Cn−1 ] ⊕ T s s [R−45 S( 1
, 1 ) Cn−1 ]
√2 √2 2 2 √2 √2

with C0 = initial line, and

Cn the Lévy C curve of order n,
⊕ splice operator, meaning add properly, i.e. T( s , s ) ,
2 2
[ means store graphics state on the GS stack and open a new one,
] means remove current graphics state oﬀ the GS stack and recall previous,
R45 means rotate US 45∘ in the PS sense,
Sa,b means scale US by a and b, in x- and y-direction
Ta,b means translate US by a and b, in x- and y-direction.
The above PS production rule transcribes systematically into the following PS def,
which has become more verbose.
!PS-Adobe-3.0 EPSF-3.0
%%Author: Kees van der Laan
%%Date: feb 2012
...
/levyC{%on stack: the order => C line
%s = size of initial segment C_0 (global)
dup 0 eq
{0 0 moveto s 0 lineto currentpoint stroke translate}
{1 sub
gsave 45 rotate .7071 dup scale levyC grestore
.5 s mul dup translate
gsave -45 rotate .7071 dup scale levyC grestore
1 add%reset order
}ifelse
}def

Systematic programming versus TACP at the expense of verbosity.

Lévy fractal as Iterated Function System ....
......
. . ..
.
.
.
.
....
..
. ...
.. .....
..
..... ... .
..... . ..
.... ...... ..
...... . .
.
.....
........

...
...
.
..
.....
....
.....
.... ...
..
..
.
..

..
..
..
....
...
...
..
.
..

..
...
.
..
.
..
...
..
.
....
......
.....
. ..
..
.
...
.....
.....
.
.
.
.
.
....
..
....
....
......
..
..
.......
....
...

....
....
..
.
..
.
..
.
..
..
..
.
...
..
.
..
.
.
.
..
.
...
..
...
.
.
.
........
......
......
..
...
...
.
...
.
.
..
.
....
........
...
....
......
.......
. ....
......
...
. .

Lauwerier(1994) in one of his exercises created a

Lévy fractal by the IFS (Iterated Function Systems)

method, which consists of 2 contracted, aﬃne

transformations, L and R, (both rotations charac-

terize Lévy) applied with equal chance.

x′ L a −b x a−1
= + and
( y′ ) (b a )(y) ( b )
x′ R c −d x 1−c
= + , a = .5, b = a = c = −d.
( y′ ) (d c )(y) ( −d )

or after substituting the parameters

x′ L 1 −1 x −1
= .5 + .5 and
( y′ ) (1 1 )(y) ( 1 )
x′ R 1 1 x 1
= .5 + .5 .
( y′ ) ( −1 1)(y) (1)

Associated with the Lévy fractal are 2 rotations with rotation centres for L: (-1,0)
and for R: (1,0) and contraction √.5 ≈ .7. Amazing, isn’t it! Laurier’s BASIC program
FRACMC2 and my conversion are given below.
Classical Math Fractals in PostScript VOORJAAR 2013 55

REM naam: FRACMC2 %!PS-Adobe-3.0 EPSF-3.0

KMAX=60000 %%Title: fracmc2
REM ***Coefficienten*** %%Author: H.A. Lauwerier(1994): Spelen met Graphics en Fractals
A=.5: B=A: C=A : D=-A %%Transcriptor: Kees van der Laan, febr 2012
DET1=A*A+B*B : DET2=C*C+D*D %%BoundingBox: -203 -54 205 206
Q=DET1/(DET1+DET2) %%BeginSetup
X=1 : Y=0 : K=0 : KMAX=10000 %%EndSetup
DO WHILE K<KMAX AND INKEYS$="" %%BeginProlog
R=RND /Courier 7 selectfont
IF R<Q THEN /x 0 def /y 0 def /halfmaxint 2 30 exp def
U=A*X-B*Y-1+A : V=B*X+A*Y+B 'rotatie L /a .5 def /b a def /c a def /d a neg def
ELSE /det1 a dup mul b dup mul add def /det2 c dup mul d dup mul add def
U=C*X-D*Y-1-C : V=D*X+C*Y-D 'rotatie R /q det1 det1 det2 add div def
ENDIF /printxy{x s y s moveto (.) show}def
X=U : Y=V %%EndProlog
PSET (X,Y) /s {100 mul}def 10 srand%10 is seed
LOOP : BEEP 10000{rand halfmaxint lt
END {/xnew a x mul b y mul sub 1 sub a add def
/y b x mul a y mul add b add def /x xnew def%rot L
}{/xnew c x mul d y mul sub 1 add c sub def
/y d x mul c y mul add d sub def /x xnew def%rot R
}ifelse
printxy}repeat
showpage
%%EOF

Helge von Koch

A von Koch broken line and a Lévy C line are related to a Lindenmayer system, also
called a rewrite system. For the von Koch broken line the rewrite is: divide a line
in 3 pieces and replace the middle piece by an equilateral triangle, with the base
omitted. Repeat the process on the 4 line pieces to the required order. It is similar to
the deﬁning construction process of the Lévy fractal; the result conveys a diﬀerent
impression, however. Below K0 … K4 , scaled with increasing order (line thickness is
scaled as well).

Properties
1. Each von Koch curve contains 4 copies of the von Koch curve of an order lower,
meaning self-similarity, which entails the production rule
Ki = Ki−1 ⊕ K60 −60
i−1 ⊕ Ki−1 ⊕ Ki−1 ,

with K0 = initial segment, ⊕ means spliced, K60 rotated over 60∘ .

2. The von Koch fractal is a historical example of a curve without a tangent. The
curve never intersects itself.
3. The length of the broken line for order n is (4⁄3)n × K0 , which with increas-
ing order n goes to ∞. The von Koch curves gave rise to the awareness that the
length of the coast of England is inﬁnite. Imagine that the yardstick has length
K0 , then all the lines above of the scaled von Koch curves have length 1! So, the
length of a fractal depends on the size of your yardstick! Awareness of grades of
inﬁnity stirred up the concept of the fractal dimension D, a jolt to the minds of
those with an iron cast idea about the 1-2-3-dimensional geometrical world. The
fractal dimension is: D = log4 ⁄ log3 ≈ 1.26.
4. Lauwerier(1987) mentions the intriguing relationship between the angle ϕk of a
segment and its index k
56 MAPS 44 Kees van der Laan

π
ϕk = ((sk + 1)mod 3 − p) with sk = ∑p−1
j=0 j
q sum of quatermals of k
3
p−1
and k = ∑j=0 qj 4j quaternary representation of k.
5. The von Koch island remains within the circumscribed circle of the initial trian-
gle (see later).
The PostScript def is an eﬃcient and concise implementation of the above speciﬁed
rewrite under property 1, neglecting scaling.
/vonKoch{%on stack order >=0; ==> von Koch
%s = size of the line segment (global)
dup 0 eq
{0 0 moveto s 0 lineto currentpoint stroke translate}
{1 sub vonKoch %lower the order on the stack and do von Koch
60 rotate vonKoch
-120 rotate vonKoch
60 rotate vonKoch
1 add %reset order
}ifelse}def

Turtle Graphics algorithm is based on the knowledge of each angle ϕk . In order to

understand, or get a feeling for, the formula mentioned, I have included a small table
for the orders 0 and 1. Order 2 yields a too long table, and has been suppressed.
order p k sk ((sk + 1)mod 3) − p ϕk
0 1 0 0 0 0
1 1 0 0 0 0
1 1 1 π⁄3
2 2 -1 −π⁄3
3 3 0 0
Lauwerier coded a BASIC program based on the knowledge of the direction of
each segment via the Turtle Graphics method. Below I have included Lauwerier’s
tiny program next to my conversion in PS.

10 REM Fractal of Helge von Koch %!PS-Adobe-3.0 EPSF-3.0

10 P=4 : DIM T(P) : PI= 3.141593 : REM***order*** %%Title: von Koch fractal a la Lauwerier
20 H=3^(-P) : PSET (0,0) %%Transcriptor: Kees van der Laan, April 2011, kisa1@xs4all.nl
30 FOR N=0 TO 4^P-1 /vonKochLauwerier{/p exch def %order => von Koch fractal (l is global)
40 M=N : FOR L=0 TO P-1 /t p array def /h l 3 p exp div def 0 0 moveto
50 T(L)=M MOD 4 : M=M\4 : NEXT L 0 1 4 p exp 1 sub{%for%n
60 S=0 : FOR K=0 TO P-1 /m exch def
70 S=S+(T(K)+1) MOD 3 - 1: NEXT K 0 1 p 1 sub{t exch m cvi 4 mod put
80 X=X+H*COS(S*PI/3) /m m cvi 4 idiv def
90 Y=Y+H*SIN(S*PI/3) }for
100 LINE -(X, Y) /s 0 def
110 NEXT N 0 1 p 1 sub{t exch get 1 add cvi 3 mod s add /s exch def}for
120 END /s s p sub def
60 s mul cos h mul 60 s mul sin h mul rlineto
Length of line-piece wired-in. }for }def
%
/l 100 def 4 vonKochLauwerier stroke
showpage
%%EOF

Note that in PS we have to convert the subscript expression for the index of an array
explicitly into integer. Another diﬀerence is that the arguments of the trigonometric
functions are in degrees in PS and in radians in BASIC.
Classical Math Fractals in PostScript VOORJAAR 2013 57

A von Koch island is a closed splicing of von Koch fractals; at right a von Koch tile
(van der Laan(1997)).
%!PS-Adobe-3.0 EPSF-3.0
%%Title: von Koch triangular island
%%...
/s 100 def
gsave .5 s mul dup neg exch translate 3 vonKochfractal pop
grestore
gsave .5 s mul dup translate
-120 rotate 3 vonKochfractal pop grestore
gsave 0 -.366 s mul translate
-240 rotate 3 vonKochfractal pop grestore
.001 setlinewidth 0 21 57.8 0 360 arc stroke
showpage
%%EOF

Lindemayer system enriched with PostScript concepts for the von Koch
fractal
What we miss in the program is the scaling to smaller size of the segments when the
order increases, as well as a more precise meaning of what spliced entails. A more
precise production rule enriched with PS concepts reads
Kn = [S 1 1 Kn−1 ] ⊕ T s 0 [S 1 1 R60 Kn−1 ] ⊕ T s s√3
3 3 3 3 3 6 6

[S 1 1 R−60 Kn−1 ] ⊕ T s −s√3 [S 1 1 Kn−1 ]

3 3 6 6 3 3

with
K0 the initial line,
Kn the von Koch curve of order n,
⊕ splice operator, meaning add properly, i.e. translate,
[ open a new GS on the GS stack,
] remove current graphics state from the GS stack and recall previous,
R60 means rotate US 60∘ in the PS sense,
Sa,b means scale US by a and b, in x- and y-direction
Ta,b means translate US by a and b, in x- and y-direction.
The above PS production rule transcribes systematically into the following PS def.

!PS-Adobe-3.0 EPSF-3.0
%%Author: Kees van der Laan
%%Date: feb 2012 order=0 order=1 order=2 order=3 order=4
...
/vonKoch{%on stack: the order => von Koch curve
%s = size of initial line segment C_0 (global)
dup 0 eq
{0 0 moveto s 0 lineto currentpoint stroke translate}
{1 sub %adjust order on the stack
gsave .3333 dup scale vonKoch grestore
.3333 s mul 0 translate
gsave 60 rotate .3333 dup scale vonKoch grestore
.1666 s mul .285 s mul translate
gsave -60 rotate .3333 dup scale vonKoch grestore
.1666 s mul -.285 s mul translate
gsave .3333 dup scale vonKoch grestore
1 add %reset order on the stack
}ifelse
}def
58 MAPS 44 Kees van der Laan

Von Koch-like fractal as Iterated Function System

Lauwerier(1994) in one of his exercises
created a von Koch-like fractal by (linear)
IFS (Iterated Function Systems), which
consists of 2 contracted, aﬃne transfor-
mations, L and R, which both for the von ....
...
. ...
....... .....
Koch fractal do contraction and mirroring. ..
. .
..
.. . ..
.
..............
. .
...
.. ......
...
..
....
.. .
..
..
..
.. ... .
.
. .. .......
For the picture at right I just took 1000 ....
...
.
........
..
.. ...
.
..
.
......
..
.
..
.

points in order to expose the dot structure, .

.....
. ... .... ........
.
. .
....... .... ..... ...... . .. ..
..... .......
. .
.... .
.
..
.. ... .... .
.. .
...
.. . ... .
. ...
.. ..
.
.. ....
......... .. .. .. .. ..
in contrast with the line structure of the
. . .......... ... . .
..... . .....
..
....... .. ....... ........
.. . .. . .. ...... .......
.... .
.......... ......

earlier approximations of the fractal.

M.F. Barnsley(1988) Fractals Everywhere, exploited contracted IFS.4
Most important property: with each contracted IFS is associated a limit ﬁgure, the
fractal attractor.

x′ L a b x a−1
= + and
( y′ ) (b −a ) ( y ) ( b )
x′ R c d x 1−c
= + , a = .5, b = .289, c = a, d = −b.
( y′ ) (d −c ) ( y ) ( −d )

or, after substitution of the parameters

x′ L .5 .289 x −.5
= + and
( y′ ) ( .289 −.5 ) ( y ) ( .289 )
x′ R .5 −.289 x .5
= +
( y′ ) ( −.289 −.5 ) ( y ) ( .289 )

Associated with the von Koch fractal are 2 rotations with mirroring with centres
for L: (-1,0) and for R: (1,0) and contraction 󰀍.52 + .2892 ≈ .58. Amazing, isn’t it!
Laurier’s BASIC program FRACMC4 and my transcription are given below. (MC is ab-
breviation for Monte Carlo, meaning alternate L and R by gambling.)
REM ***iteratief systeem, 2 spiegelingen, FRACMC4***
REM ***coefficienten***
A=.5 : B=..289 : C=A : D=-B
DET1=A*A+B*B : DET2=C*C+D*D : Q=DET1/(DET1+DET2)
X=1 : Y=0 : K=0 : KMAX=1000
DO WHILE K<KMAX AND INKEY$=" "
R=RND
IF R<Q THEN
X1=A*X+B*Y-1+A : Y1=B*X-A*Y+B 'spiegeling L
ELSE
X1=C*X+D*Y+1-C : Y1=D*X-C*Y-D 'spiegeling R
END IF
X=X1 : Y=Y1
PSET (X,Y),10
K=K+1
LOOP : BEEP
END
Other values of the parameters
a=.5 b=.5 c=.6667 d=0 %bebladerde tak
a=.5 b=.289 c=.5 d=-.289 %von Koch
a=.5 b=.5 c=.5 d=0 %kale tak
a=.5 b=.5 c=.6 d=-.2
a=0 b=.64 c=0 d=-.64 %tegelpatroon
Classical Math Fractals in PostScript VOORJAAR 2013 59

%!PS-Adobe-3.0 EPSF-3.0
%%Title: fracmc4
%%Author: H.A. Lauwerier(1994): Spelen met Graphics en Fractals
%%Transcriptor: Kees van der Laan, febr 2012
%%BoundingBox: -100 -1 103 60
%%BeginSetup
%%EndSetup
%%BeginProlog
/Courier 7 selectfont
/x 1 def /y 0 def /halfmaxint 2 30 exp def/maxint 2 31 exp 1 sub
def
/a .5 def /b .289 def /c a def /d b neg def
/det1 a dup mul b dup mul add def /det2 c dup mul d dup mul add
def
/q det1 det1 det2 add div def
/printxy {x s y s moveto (.) show}def
%%EndProlog
10 srand%10 is seed
/s {100 mul }def%scaling
1000{rand maxint div q lt
{/xnew a x mul b y mul add 1 sub a add def
/y b x mul a y mul sub b add def /x xnew def%mirror L
}
{/xnew c x mul d y mul add 1 add c sub def
/y d x mul c y mul sub d sub def /x xnew def%mirror R
}ifelse
printxy
}repeat
showpage
%%EOF

Deterministic von Koch and randomness

Peitgen c.s.(2004) mentions the deterministic von Koch fractal combined with ran-
domness, and states that a better model for coastlines is obtained.

%!PS-Adobe-3.0 EPSF-3.0
%%Title: von Koch Random fractal, 2012
%%Author: Kees van der Laan, kisa1@xs4all.nl
%%BoundingBox: -5 -125 650 15
%%BeginSetup
%%EndSetup
%%BeginProlog
/vonKoch{dup 0 eq
{0 0 moveto s 0 lineto currentpoint stroke translate}
{1 sub vonKoch
pm{ 60 rotate vonKoch
-120 rotate vonKoch
60 rotate vonKoch}
{ -60 rotate vonKoch
120 rotate vonKoch
-60 rotate vonKoch}ifelse
1 add
}ifelse
}def
22121943 srand
/pm{rand 1073741823 gt{true}{false}ifelse}def
%%EndProlog
%
%Program ---the script---
/s 5 def % s = initial size of line piece
1 vonKoch pop % order 1
2 s mul 0 translate 2 vonKoch pop
3 s mul 0 translate 3 vonKoch pop
showpage
60 MAPS 44 Kees van der Laan

KRONKEL is Lauwerier’s universal program to construct fractal islands based on

similarity transformations.

10 REM ***KRONKEL: Fractal polygonal Island and %!PS-Adobe-3.0 EPSF-3.0

model*** %%Titel: Kronkel ---Koch Island--- H.A. Lauwerier
10 DIM x(4096), Y(4096) %%Transcriptor: Kees van der Laan, kisa1@xs4all.nl, April 2011
20 U=4 : DIM A(U), B(U) : REM ***Number of sides of Island*** /kronkel%order p ==> fractal island (globals s, u, a, b, v , c, d)
30 V=4 : DIM C(V), D(V) : REM ***Number of pieces of model*** {kronkeldict begin%push kronkeldict on the d-stack
40 DATA 1,1,-1,1,-1,1,-1,-1,1,-1,1,1 : REM ***Corners island*** /p exch def
50 DATA .3333,0,.5,.2887,.667,0 : REM ***Data model*** /x u v p exp mul cvi array def /y u v p exp mul cvi array
60 INPUT P : REM ***Choice order*** def%auxiliaries
70 FOR I=0 TO U : READ A(I), B(I) : NEXT I %calculate coordinates corners `kronkel'
80 FOR I=1 TO V-1 : READ C(I), D(I) : NEXT I x 0 0 put y 0 0 put
90 REM ***Calculation coordinates Kronkel line*** x v p exp cvi 1 put y v p exp cvi 0 put
100 C(0)=0 : D(0)=0 : x(0)=0 : Y(0)=0 : X(v^P)=1 : Y(V^P)=0 0 1 p 1 sub{/i exch def %for i
110 FOR I=0 TO P-1 0 v p i sub exp v p exp 1 sub{/j exch def %for j
120 FOR J=0 TO V^P-1 STEP V^(P-I) /m1 j v p i sub exp add def
130 M1=J+V^(P-I) : DX=x(M1)-X(J) : DY=Y(M1)-Y(J) /dx x m1 cvi get x j cvi get sub def
140 FOR K=1 TO V-1 /dy y m1 cvi get y j cvi get sub def
150 M2=J+K*V^(P-I-1) 1 1 v 1 sub{/k exch def %for k
160 x(M2)=DX*C(K)-DY*D(K)+X(J) /m2 j k v p i sub 1 sub exp mul add def
170 Y(M2)=DY*C(K)+DX*D(K)+Y(J) x m2 cvi dx c k get mul dy d k get mul sub x j cvi get add puty m2
180 NEXT K cvi dy c k get mul dx d k
190 NEXT J get mul add y j cvi get add put }for%k
200 NEXT I }for%j
210 REM ***DRAW ISLAND*** }for%i
220 PSET(A(0),B(0)) %create path for each side m of base line
230 FOR M=0 TO U-1 a 0 get b 0 get moveto
240 DA=A(M+1)-A(M) 0 1 u 1 sub{/m exch def%for m
250 DB=B(M+1)-B(M) /da a m 1 add cvi get a m cvi get sub def
260 FOR N=0 TO V^P /db b m 1 add cvi get b m cvi get sub def
270 LINE -(DA*X(N)-DB*Y(N)+A(M), DB*X(N)+DA*Y(N)+B(M)) 0 1 v p exp{/n exch def%for n
280 NEXT N da x n cvi get mul db y n cvi get mul sub a m cvi get add
290 NEXT M db x n cvi get mul da y n cvi get mul add b m cvi get add lineto
300 END }for%n
}for%m
Length of size of line-piece is wired-in. end}def %end pops kronkeldict off the d-stack
/kronkeldict 8 dict def
%
/s {50 mul} def %scale
In PS /u 4 def %number of corners of the island
/a [ 1 s -1 s -1 s 1 s 1 s ] def %x coordinates of corner points
s scaling
/b [ 1 s 1 s -1 s -1 s 1 s ] def %y coordinates of corner points
u a b size and corners of island /v 4 def %number of line pieces of the model line
v c d size and corners of model line /c [ 0 .3333 .5 .6667 ] def%x coordinates of corners
have been used as globals, and could have been /d [ 0 0 .2887 0 ] def%y coordinates of corners
initialized within the dictionary 2 kronkel stroke showpage
%%EOF
No ﬁxed bounds on x and y

order=0 order=1 order=2 order=3 order=4

inward islands
His islands are based on similarity transformations, not on the calculation of the
direction of the next line piece as in the line fractals. The Kronkel program can
also be used for degenerated islands, i.e. for line fractals, such as Lévy, von Koch,
Minkowski, …
The order of specifying the corners of the island determines whether the fractal is
drawn inside (anti-clockwise speciﬁcation) or outside (clockwise speciﬁcation)
Classical Math Fractals in PostScript VOORJAAR 2013 61

order=0 order=1 order=2 order=3 order=4

outward islands
...
(C:\\PSlib\\PSlib.eps) run
%globals s, u, a, b, v, c, d
/s {30 mul} def %scale
/u 4 def
/a [-1 s -1 s 1 s 1 s -1 s] def %abcissae corners island (clockwise=>inside, scaled)
/b [-1 s 1 s 1 s -1 s -1 s] def %ordinates corners island (clockwise=>inside, scaled)
/v 4 def
/c [0 .3333 .5 .6667 ] def %abcissae corners broken model line
/d [0 0 .2887 0 ] def %ordinates corners broken model line
0 kronkel stroke gsave
110 0 translate 1 kronkel stroke grestore
showpage

A triangular island (0, 0), (1, 0) (.5, .866) (0, 0) can equally-well be speciﬁed, with the
broken model line (0, 0), (.5, 0), (.375, .2165), (.5, 0), (.625, .2165), (.5, 0).

order=0 order=1 order=2 order=3 order=4

The degenerate Lévy island can be speciﬁed by the line (-1, 0), (1, 0), with the
(broken) model line (0, 0), (.5, 0)

%...
(C:\\PSlib\\PSlib.eps) run
%globals s, u, a, b, v, c, d order=0 order=1 order=2 order=3 order=4
/s {30 mul} def
/u 1 def /a [ -1 s 1 s ] def%abcissae corners line (scaled)
/b [ 0 0 ] def%ordinates corners line (scaled)
/v 2 def /c [ 0 .5] def %abcissae corners broken model line
/d [ 0 0 ] def %ordinates corners broken model line
0 kronkel stroke
110 0 translate 1 kronkel stroke
%...
showpage
%%EOF

An interesting program to experiment with. The PS transcription is also included in

my PSlib. Lauwerier provides moreover variants: KRONKELT, biased by the number
system (Dutch talstelsel) approach and KRONKELB, where backtracking has been used.

Minkowski fractal
Much similar to the von Koch fractal is the Minkowski fractal, called sausage by
Mandelbrot. The replacement scheme can be distilled from the illustration below,
especially M0 → M1 .
10 REM ***Sausage of Minkowski***
10 DIM A(7) : A(0)=0 : A(1)=1 : A(2)=0 : A(3)=3
20 A(4)=3 : A(5)=0 : A(6)=1 : A(7)=0
30 P=3 : DIM T(P) : REM***order***
20 H=4^(-P) : X=0 : y=0: PSET (0,0)
30 FOR N=0 TO 8^P-1
40 M=N : FOR L=0 TO P-1
50 T(L)=M MOD 8 : M=M\8 : NEXT L
62 MAPS 44 Kees van der Laan

60 S=0 : FOR K=0 TO P-1

70 S=(S+A(T(K))) MOD 4: NEXT K
80 IF S=0 THEN X=X+H
81 IF S=1 THEN Y=Y+H
82 IF S=2 THEN X=X-H
83 IF S=3 THEN y=y-H
90 LINE -(X, Y)
91 NEXT N
92 END

Length of line-piece is wired-in.

%!PS-Adobe-3.0 EPSF-3.0
%%Title: Minkowski sausage by H.A Lauwerier
%%Transcriptor: Kees van der Laan, kisa1@xs4all.nl,m April 2011
/mink{/p exch def%order, global l length of initial line
/a [ 0 1 0 3 3 0 1 0] def%model specification by direction numbers
/t p array def 0 0 moveto
/h l 4 p exp div def
0 1 8 p exp 1 sub{/m exch def
0 1 p 1 sub{t exch m cvi 8 mod put /m m cvi 8 idiv def}for
/s 0 def
0 1 p 1 sub{/s a t 4 -1 roll get get s add cvi 4 mod def}for
s 0 eq { h 0 rlineto}if
s 1 eq { 0 h rlineto}if
s 2 eq { h neg 0 rlineto}if
s 3 eq { 0 h neg rlineto}if
}for stroke} def
%
/l 100 def 3 mink showpage
%%EOF

0 1 order=2 order=3 order=4

log 8
The fractal dimension of the Minkowski fractal D = log(1/4 −1 ) = 1.5. The array a

contains the direction numbers: 0, 1, 3, meaning direction 0 , 90∘ , -90∘ , respectively.

∘

Minkowski island The essentials of the island program in PS are given below.
%!PS-Adobe-3.0 EPSF-3.0
%%Title: Minkowski island
%%...
/l 200 def 4 mink
gsave l 0 translate -90 rotate 4 mink grestore
gsave l l neg translate -180 rotate 4 mink grestore
0 l neg translate -270 rotate 4 mink
showpage
%%EOF

Dragon ﬁgures
1 1 1 1
-1
1 1 -1 -1 1
-1 -1 1 -1

Folding a strip of paper repeatedly and after un- -1

-1

1
1
1
-1 -1 -1

1
1

folding one may ask: How to draw the meander? 1

The curve with rounded 90∘ corners is named Dragon curve by Heighway. The
curve does not intersect itself. A nice example for developing the mathematical prob-
lem solving attitude in discovering the intriguing pattern. (Be aware of folding con-
sistently in the right direction.)
Classical Math Fractals in PostScript VOORJAAR 2013 63

Let us set up a table, where for each line piece the continuation angle is given: r
means rotate −90∘ , and l means rotate 90∘ , and unearth the regularity in the direc-
tions d(n), for n = 1, 2, 3, …
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32
r r l r r l l r r r l l r l l r r r l r r l l l r r l l r l l r
d(n) = r(igℎt) for n = 1, 5, 9, … d(n) = l(eft) for n = 3, 7, 11, … d(n) = d(n⁄2)
for n is even.
Express n in the form k × 2m where k is an odd number. The direction of the nth
turn:
if k mod 4 = 1 then the nth turn is r;
if k mod 4 = 3 then the nth turn is l.
The direction of turn 76376: 76376 = 9547 × 8 & 9547 mod 4 = 3 → d(76376) = l.

10 REM Draak %!PS-Adobe-3.0 EPSF-3.0

10 P=3 : REM***order*** %%Title: Draak by H.A Lauwerier
20 H=2^(-P/2) : A=1.7453 : REM ***Corner*** %%Transcriptor: Kees van der Laan, kisa1@xs4all.nl, April 2011
30 B=PI-A : X=H : Y=0 : LINE (0,0)-(H,0) : S=0 /draak{%draak curve with angle A and order p, global scaling s
40 FOR N=1 TO 2^P-1 : M=N %angle p ==> dragon curve
50 IF M MOD 2 = 0 THEN M=M/2: GOTO 50 {/p exch def /b 180 3 -1 roll sub def%order p and angle b
60 IF M MOD 4 =1 THEN D=1 ELSE D=-1 /h 2 p -2 div exp s def %size piece scaled
70 S=S+D 0 0 moveto h 0 lineto
80 X=X+H*COS(S*B) 1 1 2 p exp 1 sub{%for n
90 Y=Y+H*SIN(S*B) : LINE -(X,Y) /m exch def
91 NEXT N {m cvi 2 mod 0 eq {/m m 2 div def}{exit}ifelse}loop
92 END m cvi 4 mod 1 eq {/d b neg def}{/d b def}ifelse
d rotate h 0 rlineto%no currentpoint translate necessary
}for %n
}bind def %in library for local variables with draakdict
%
/s {30 mul} def 100 3 draak stroke showpage
%%EOF

For the order p =14 and angle 90∘ I reproduced Lauwerier’s result in PS, see below
at left.

The number of line pieces is 2p . The curve of order 10 with rounded corners is at
right. The curves don’t intersect themselves, which is seen in the figure with rounded
corners. (In Lauwerier’s program the direction D is not in agreement with the folded
paper and the dragon figure. This is adapted in the PS code.)
Knuth in the TEXbook Appendix D p390 also mentions the dragon curve in rela-
tion to Turtle Graphics, and draws dragon figures in TEX. When I tried the order 12
in TEX, in 1995, TEX gave the error message ‘TEX capacity exceeded.’
64 MAPS 44 Kees van der Laan

Dimensions The bounding box obeys the pro-

portion 3 : 2. The fractal dimension of the
curve equals 2, a local plane ﬁlling curve ....
..
.
..
....
..
.
.
..
.
.
..
..
.
..
. .
..
..
..
..
..
.
..
.....
.
.
.
.
.
....

(Courtesy http://en.wikipedia.org/wiki
..
..
...
...
. .. .
..
...
..
...
...
...
. ..
..
. ..
....
....
...
..
.
..
. .
..
...
..
..
.
.
..
.
..
..
.. .
..
.
.
..
.
. .
....
..
..
. ......
....
...
..
...
...
.
...
....
.. ....
...
. ..
....
..
.
...
..
.
..
.
.....
.
..
..
.....
..
. ..
.... .. .. .. ..
..
...
. .
........ ..
...
..
. . ..
....
....
....
..
.. ....
.
.. ..
....
.. ..
.
..
...
..
....
..
.
..
.. ... ..
. .
.
.
.
..
..
..
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
.
.
.
.
.
.
.
..
.
..
. ..
...
....
.....
..
.
.
..
..
...
.
.... ..
.
.
..
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
..
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
.
..
.
.
.
.
..
.
.
.
.
...
.
.. ..
..
. ... .... .
..
.
.
...
..
.
..
.
..
....
.
.
..
.
.
.
..
.
.
.
..
.
.
.
..
.
.
.
..
.
.
.
..
.
.
.
..
.
.
.
..
.
.
.
....
..
.
.
...
.. ..
...
.
.
.
..
...
..
..............
..
. .. .

/Dragon_curve, which mentions more proper-

....
. ..
.. ....
.
....
...
...
.....
.. .. ..
..
...
. ..
.
..
...
. ..
....
..
...
...
...
...
...
...
...
...
...
...
..
. ..
.
....
. ..
....
. .....
..
. ..
..
. ...
.. ..
.
...
.
...
..
....
.
...
.
...
.
...
.
...
.
...
.
...
.
...
.
...
...
.
...
..
... ..
.
...
..
.
..
.. .
...
..
..
..
. ..
....
... .. ...
..
. ..
.....
..
..
...
..
...
....
..
..
....
. ..
..
. .
.
...
..
.
..
...
..
...
..
..
..
..
..
...
..
..
..
..
..
..
.
. .
..
...
...
..
.. ..
.
..
. ..
..
. .
..
...
. ..
..
..
. ...
.....
..
............. ..
..
.........
.
..
. ..
....
...
...
..
..
.
....
..
..
. . .... . ...
..
..
..
..
.
..
.. .
..
...
...
...
..
.
..
..
..
..
..
......
..
..
....
...
..
. .
. ...
. ..
..
.
..
..
..
..
..
..
..
..
..
..
..
..
..
..
..
..
..
..
..
..
..
..
...
....
..
..
.......
.. .. ..
. ...
...
.
...
.
..
...
.
...
.
. ..
.
..
.
..
...
...
...... ..
.
..
..
...
...
....
..
.
...
.
...
.
...
.
...
.
...
...
...
..
...
..
...
..
...
.
...
..
...
..
...
.
...
...
.
...
.
.. ..
..
....
.
...
.
...
.
...
.
...
.
...
.
...
....
..
.
..
...
..
.
...
.
....
..
...
. ..
..
..
..
...
..
...
....
..
.
...
.
...
.
..
.
.
..
.
...
.
...
.
...
.
...
.
...
.
...
.
...
.
...
.
...
.
...
.
...
.
...
.
...
.
...
.
...
.
...
.
...
.
...
.
...
.
...
.
...
.
...
.
..
...
..
..
...
..
. ...
.. .
.
....
. ...
..
..
. ..
....
..
.. .
.
...
..
.
..
.
..
.
..
.
..
.
...
.
...
. .
..
..
.
..
.
..
. ..
.
..... ..
.. ..
...
..
..
..
..
..
..
..
.
..
.
..
.
..
.
..
.
..
..
..
..
.
..
..
.
..
.
..
.
..
.
..
.
..
..
..
.
..
.
..
.
..
.
..
. .
.
..
..
.
..
.
..
.
..
.
..
.
..
..
..
..
.
..
..
.
..
.
..
.
..
.
..
..
..
..
.
..
..
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
..
.
..
.
..
..
.
..
..
.
..
.
...
..
..
. .
.. ...
. . ..
..
...
.
. ..
.....
...
..
....
..
...
. ...
..
.
. ....
..
.
....
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
...
.
..
..
..
.
..
.
..
.
...
.
...
. ....
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
..
.
..
.
..
....
..
.
..
.
..
.
..
.
..
.
..
.
..
..
..
..
..
..
.
..
.
..
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
..
..
.
..
.
...
.
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
....
....
...
...
...
...
...
...
.. ..... ....... ..
.. ...
...
..
..
...
..
..
. .
..
..
..
..
. ..
..
... .
...
...
...
...
....
..
.. ..
...
. ...
...
...
..
..
. .
...
...
...
...
...
...
....
..
.. .
...
...
...
..
. .
...
...
...
...
...
...
...
..
..
. ..
....
...
...
..
...
...
...
...
...
.....
........ .. ...
...............
... ...
...
...
...
...
...
...
...
...
...
...
...
...
...
....
..
.. .
...
...
...
...
...
.....
...
...
..
..
..
..
..
..
..
..
..
..
..
..
..
..
..
..
..
..
..
..
..
..
...
...
..
..
...
. ....
.. ..
..
.
...
...
.
.
.. ....
..
.
..
.....
..
...
..
..
..
...
..
...
...
..
. .
..
...
...
...
...
...
..
....
....
....
....
...
..
...
...
..
. ..
....
...
...
...
...
...
. .. ...
..
. .
...
. ...
....
..
...
. ..
. . ... .. . ..
...
..
..
..
.
.
...
...
. ..
....
..
.
..
.
.
.
.
...
...
...
...
...
..
.
.
...
.
.
.
..
..
...
.
.
..
..
.
..
..
.
...
. ..
..
....
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
.
.
.
..
...
..
...
..
.
..
.
..
..
..
.
.
..
...
..
..
.
..
..
...
.
.
.. . .
....
. ...
.
.
.
..
..
..
..
..
..
...
.. ..
. ..
.... ..
..
...
.. . .
....
..
..
.... ..
...
........
...
. ..
. .
..
....
...
. ... .. ..
... ...
.... . . .
....
... .
.... .. .
..
..
...
. .
. ..
...
....
...
..........
..
..
.....
...
..
...
...
..
......
..
.
..
..
..
....
..
......
...
.
..
..
.
..
..
.
..
.
..
.
..
.
..
..
..
. .
.
..
..
..
.
. .
.

ties of the Dragon curve, such as its self-simi-

... . ..... . . . .. .. . . .. ... . ... ..
. . . ..................... ........... . ......... . . .... .... .........
..
....
..
...
....
..
.
...
.
..
...
..
...
.
...
.
...
... .. ..
.
...
.
...
.
....
.
. .... ..
......
....
.
...
..
..
.
.
.
..
..
..
.
..
...
..
...
..
...
..
..
..
...
..
...
.
..
...
..
...
.
...
.
...
.
...
.
...
.
...
...
.
..
.... ..
.
. ..
...
.
...
.
...
.
...
.
... ..
..
...
..
....
.
..
..
...
..
.
..
...
..
..
...
..
.
...
.
...
.
. ..
..
..
...
.
...
.
..
.
.
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
.
.
.
.
.
.
.
.
.
.
..
.
.
...
.
..
.
.
.
.
.
..
.
..
.
.
..
.
.
..
.
..
.
.
...
...
..
.. .. ..
. .
.
.
.
.
.
..
.
.
.
.
.
.
.
.
..
.
.
.
..
..
.
.
.
.
.
.
.
.
..
.
..
.
..
.
..
.
..
.
..
..
.
..
..
..
..
..
.
..
..
..
.
..
.
..
..
..
.
..
..
..
..
..
..
.... .
. .
..
..
..
..
..
..
.
..
.....
...
.
..
..
. .
..
..
..
..
.
..
..
..
.
..
.
..
.
..
.
..
.
..
.
..
.
.. .
. ..
..
..
.... ....
.....
..
.
..
.
..
.
..
.
..
.
..
.
..
..
.
..
.
..
.
..
.
..
.
..
.
..
..
.
..
.
..
.
..
.
..
..
..
. .
... .. .
.
..
..
.
..
.
..
.
.. ..
..
...
. ..
..
...... ..
.
...
..
. .
........
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
..
.
..
.
..
.
...
.
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
...
....
...
...
...
...
...
....
. ....
..
..
..
...
.
..
.
..
..
..
..
..
..
.....
.
..
..
..
..
..
.
..
.
..
.
..
.
.
..
.
..
..
.
..
.
..
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
..
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.....
.
.
....
..
.
...
..
...
... . ....
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
.. ..
.
..
. ..
...
... ..
.
..
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
..
.
..
.
..
.
..
..
...
. ..
..
..
.
..
.
..
. ...
.
...
..
......
...
..
.... ..
..
..
...
. ..
...
..
.
..
.
..
.
..
..
.
...
.
..
.
..
.
..
..
.
..
.
..
.
..
.
..
.
..
..
.
..
.
..
.
..
.
..
..
.
..
.
..
.
..
.
..
..
.
..
.
. .
...
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
....
. .
.
.
..
..
..
.
.
.
.
.
.
.
.
..
..
.
.
.
.
.
.
.
.
..
..
..
..
.
.
.
.
...
..
...
...
...
...
. .
.
...
.
..
.
.
..
..
. .
.
..
..
..
.
..
..
.
..
...
.
..
.
..
..
..
.
..
.
..
...
.
..
.
..
.
..
.
..
.
..
.
..
.
..
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
... .
......
..
..
. .
..
.....
...
..
.. .
..
..
. .
...
..
. .
..
. .
..
. .
..
...
. .
.. ...
. .. ..
........ . .. .
.....
..
. .
..
. .
..
. .
..
. .
..
. .
..
. ..
...
. .
..
. .
..
. .
..
. ..
...
. .
..
. .
..
. . .
....
. ..
...
.. .
....
..
. .
..
. .
..
....
. ...
...
..
. .
..
. .
..
... ..
. ...
...
...
...
... .
...
.....
.....
...
...
....
.....
....
.....
.....
...
...
...
...
..
. ..
..
..
..
..
..
..
..
..
..
.. ..
..
..
..
.. .. .
....
..
..
..
..
..
...
..
...
. ..
...........
.. ..
..
. .
..
...
..
..
..
...
.
..
.
..
.
..
.
...
.
..
..
.
..
..
.
..
.
...
.
..
.
..
.
..
.
..
..
..
.
..
.
..
.
..
.
..
.
..
.
..
..
.
..
.
..
.
..
.
..
.
..
..
..
..
.
...
..
.... .
..
..
. .
..
..
..
..
..
..
. ...
..
. ..
..........
... ...
.. ..
...
.........
. ....
... ..
. . ...
.....
..... .. .
..
....
. ..
.......
..
..
..
..
. .
..
. .
..
..
....
..
..
..
..
.
..
.
...
.
..
..
..
.
..
.
..
.
..
.
..
..
.
..
.
..
.
..
.
..
..
...
..
..
.
..
.
...
..
. ..
.... .....
..
.
..
.
..
.
..
..
...
. ..
...
. .
.
..
..
..
.
..
.
..
..
.
..
...
.
.
..
.
.
..
.
..
..
..
.
..
.
..
.
..
..
..
.
..
.
..
.
..
.
..
.
..
.
..
..
.
..
..
..
.
..
.
..
.
..
.
.
.. .....
...
.....
. ..
..
...
. ...... ... ... ..
... .
. ...
...
..
....
..
.
.
..
..
..
.
.
.
.
..
..
.
.
.
.
.
.
..
..
...
.
.
.
.
.
.
..
..
...
.
.
.
.
.
.
..
..
..
...
.
.
.
...
...
...
.
...
..
. ...
..
..
.
.
..
.
.
.
.. ...
...
..
.
.
. .
..
.
..
..
.
.
.
.
..
.
.
.
.
.
..
..
.
.
.
..
.
.
...
. .
..
.
.
...
.
..
.
..
..
.
..
..
.
..
.
.
..
.
..
.
.
..
..
..
..
.
.
..
.
..
.
.
..
..
..
..
...
. ..
..
..
. ..
.
.
.
.
.
..
.
.
.
.
.
.
.
.
.
..
.
.
.
.
.
.
.
.
...
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
. . ..
..
..
... ..
.. ..
..
..
..
.... ....
......
...
...
...
.. .
...
..
........
....... .. . .
...
..
..
...
....
..
....
...
...
...
...
..
..
..
..
.....
...
..
..
..
..
.....
...
.. ..
..
...
..
.. .
....
..
...
....
.. . ..
..
..
..
...
...
..
...
..
..
...
..
..
..
...
..
..
...
..
..
..
..
. ..
..
. ..
.. .
.
...
.
..
.
...
.
..
...
.
...
.
....
.... .
.....
.
...
.
...
.
..
.
.. ........ . . .
.... .
..
.
.....
.
..
.
.....
.
..
.
....
.....
.
....
..... .
.
. . ..
...
..
..
... .
.
...
...
.
..
..
..
..
..
..
..
...
..
.
..
..
..
..
... .... ..
......... ..............
..
. .
.
....
. ...
..
....... ..
..
. ......... ............
....
..
.. .
...
...
...
....
..
.. .
...
....
..
....
..
..
..
..
.. ..
....
..
.. .. ..
..
.. ..
..
...
..
....
..
...
...
...
..
.. ..
..
..
..
..
..
..
..
..
..
..
..
..
..
..
...
..
..
..
..
..
..
...
. .. .
.....
. .
.
..
..
..
..
...
..
..
..
..
..
..
.
..
. .
.... ....
.. .

larity and the spiral shape.)

...
. ..
..
..
...
..
..
..
..
..
..
..
. .
..
..
..
..
..
..
..
..
..
..
..
..
..
.....
.. .....
..
. . ..
...
.
....
..
..
.. ...
.. ....
..
..
..
..
..
..
..
..
..
...
..
..
..
..
..
..
..
..
..
..
..
................... ...... .
... . ..... .
..
..
..
...
..
..
..
..
..
..
..
..
..
..
..
...
..
..
..
..
. .
..
. .
..
. ..
..
..
..
..
..
..
..
..
... ... ..
..
..
..
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
..
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
..
.
..
.
..
.
..
.
..
. ...
..
..
.. .
.. .
..
...
.....
..
...
..
..
..
..
.
..
.
..
.
..
...
. ... ..
.
..
.
..
..
.
..
.
..
.
..
.
..
.
..
.
..
..
.
..
.
..
..
.
..
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
.
..
..
.
..
.
..
.
..
.
..
..
.
..
.
..
.
..
.
..
..
.
..
.
..
.
..
.
..
.
..
.
..
.. ..
..
..
..
.
..
.
..
. .
.
..
.
..
..
.
..
.
..
. .
.
..
.
..
.
..
.
..
.. ..
..
..
. .
..
.
..
.
..
.
..
.
..
..
.
..
.
..
.
..
.
..
..
.
..
.
..
..
.
..
.
..
.
. .. .
.. .
..
..
.. ....
...
.. .
..
..
.
..
.
..
.
..
.
..
.
..
..
...... ....
..
.
..
.
.
..
.
....
...
..
... ..
.
..
...
...
...
...
...
...
...
...
...
...
...
. ..
..
...
...
...
. ..
...
...
...
...
.....
. ...
. ..
...
....
. ...
..
.
. ..
...
...
. ..
....
..
...
...
...
.....
.
..
...
...
...
...
...
...
...
.....
. .. ..
....
. ...
...
...
...
...
...
..
...
...
.....
..
..
...
. ..
..
.
...
...
...
. ..
..
.
...
...
...
. ..
..
...
....
. ..
...
...
...
.. ..
..
..
.
. ...
.
....
.
...
.....
..
..
..
. ...
..
..
..
..
...
.......
...
..
.. . ..
..
..
..
.
.. .......
...
.....
...
..
....
...
..
..
..
..
.. .. .
.....
. .
..
... .
. ...
..
..
..
..
..
.. .
...
...
..
...
..
.
...
..
..
.
..
.
..
..
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
...
.
...
.
.
..
.
.
.
.
.
.
.
.
..
.
.
..
...
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
..
.....
.
..
.
.
.
.
.
.
.
.
.
...
.
.
.
.
.
..
.
.
..
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
..
..
.
..
.
.
.
..
.
.
.
.
.
.
.
.
.
.
..
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
.
.... ..
..
..
...
...
..
.
..
.
.
.
.
.
.
.
.
.
.
.
.
...
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
. .
..
..
.
.
.
.
.
.
.
.
..
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
.
.
.
.
.
..
.
.
.
..
.
.
.
.
.
.
.
.
.
.
.
..
.
..
.
.
.
....
..
....
.
...
.
..
...
..
.
..
.
..
..
....
.
...
.
....
.
..
. ..
..
.
...
..
.. ...
. .
.
. ...
. .
..
...
. . .
.
..
.
.
..
.
..
....
.
.
..
... . ..
.
..
..
..
....
.
..
..
.
.
..
..
.
.
..
.
.
.
..
..
..
....
.
.
..
.
.
.
..
.
..
.
..
..
.
.
..
..
.
.
..
.
.
.
..
.
.
.
..
.
.
.
..
.
.
.
..
.
.
.
..
.
..
.
....
..... ....
..
.
.
.
.
.
.
..
.
.
.
. .
.. ..
..
..
.
.
..
.
.
.
..
.
.
.
..
.
.
.
..
.
.
.
..
.
.
.
..
.
..
.
..
....
.
...
.
...
...
.
....
....
. ....
..
.. ..
... ....
..
..
..
..
...
. .....
..
..
..
..
..
..
...
.
...
..
...
..
..
..
..
..
..
..
..
..
..
...
..
..
..
..
..
..
...
. .. ..
.. .
..
..
..
..
..
..
..
..
..
..
..
..
...
..
..
..
..
..
..
..
..
...
..
..
..
..
..
....
..
..
..
..
..
..
..
..
..
. .
.. ..
..
...
.. . ..
...
....
..
..
...
..
..
...
..
..
..
..
..
..
..
......
...
. ..
..... . ... ....... ....
..
....
....
.. .
..
.. .
...... .. ...
....
..
.....
..
.... .
..
. ....
.....
. .
..
. ..
....
...
....
..
..
. .
...
...
...
...
.....
.. . ..
.....
...
. .
....
...
. .
..
.......
....
...
. .
..
. .
..
. .
..
. .
......
.....
. ..
..
..
. .
...
...
...
...
...
...
....
..
.. .
..
.....
.
..
.
...
..
..
..
..
..
.
...
..
.
..
.
..
.
..
..
...
...
..
.
..
.
..
.
.
.. ..
.
..
..
..
...
.. ...
..
.
..
.
..
....
..
.
..
.
..
.
..
..
.
..
.
..
..
.
.
..
.
..
.
....
..
.
..
.
..
..
..
..
..
.
..
.
..
..
. .
..
..
..
....
...
..
.. ..
. ..
....
. ..
...
.....
..
.. .
..
. .. ..
....
.
. .
....
..
.. .. .
. ...
. ...
..
.
...
.
..
...
.
...
..
..
..
...
...
..
..
... .
..
...
.
...
..
.. .
.
...
..
..
.. .
..
.
..
.. .
....
... ...
.... ... ...... ......
...
.
..
..
.
...
...
... .
.. ..
.. . .
. . .
.
.. .
..
.
..
. ..
.
..
..
...
......
.
..
.
..
.
..
....
.. .
..
..
...
.
..
........
. . ...
...
...
. ...
....
. ..
.. .
..
....
...
.
.
.
.
.
.
.
...
.
..
..
.
.
..
.
..
.
.
.
.
.
..
.
.
.
.
.
...
.
...
.
.
..
.
.
.
.
.
..
.
.
.
.
...
.
...
....
..
..
.
.
..
.
.
.
..
.
.
..
.
.
..
.
.
..
.
.
..
.
.
..
.
.
..
.
..
.
.
.
.
.
.
.
.
.
..
.
..
.
..
..
.
.
.
.
.
..
.
.
.
.
.
..
..
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
..
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
.
.
.
.
.
.
.
.. ..
..
..
.
.
..
.
..
..
..
.. .
.
..
...
..
..
..
..
..
..
..
.
..
.
..
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
..
.... ...
..
.
..
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
.
..
.
.
.
.
.
.
.
.
.
.
..
.
.
.
..
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
.
..
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
.
.
.
.
.
.
.
..
.
.
..
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
.
.
..
.
.
..
.
..
.
.
.
..
.
.
..
.
.
.
.
.
.
.
.
.
.
..
.
.
.
..
.
.
..
...
.
. ...
..
..
...
...
...
.....
..
. ..
..
. ... .........
..
...
. ..
..
. ..
..
...
...
...
...
.....
..
...
. ..
..
...
...
...
....
..
...
...
...
...
...
...
...
...
..
..
....
..
..
...
..
..
..
..
..
..
....
..
..
...
..
........
...
..
....
.. ...
..
..
..
..
..
..
..
..
..
..
..
..... ......
..... ..... .. .
.. ....
... ..
.... .....
..
.....
...
.....
.......... . ...
. .
...
.. .

At right a ﬁlled dragon-like Julia set.(More on

Julia fractals: JULIA fractals in PostScript, sub-

mitted for MAPS.)

The Dragon curve can be generated similarly to the rewriting scheme of the Lévy
fractal, with parts rewritten mirrored.

Star fractals
As introduction a generalization of the program star of the Blue Book p51. The pro-
gram is more general because it allows to draw the pentagram or the 5-star depend-
ing on the value of the angle parameter. Moreover, the number of vertices can be
varied, to obtain for example a heptagon casu quo 7-star (heptagram).
/gonstar%p (order) v ==> star
{gonstardict begin /v exch def /angle exch def
0 0 moveto
v{angle rotate l 0 rlineto}repeat closepath
end} bind def
/gonstardict 2 dict def
%%EndProlog
%
%Program ---the script---
%
/l 100 def 144 5 gonstar stroke
gsave 75 -25 translate
/l 50 def 1.415 setmiterlimit
72 5 gonstar stroke grestore
gsave 0 -110 translate
/l 100 def 1080 7 div 7 gonstar stroke
grestore
gsave 65 -130 translate
/l 35 def 1.415 setmiterlimit
360 7 div 7 gonstar stroke
showpage

Lauwerier’s ingenious, concisely programmed star fractal illustrations, left and right
below, consist also of 1 (broken) line.
Classical Math Fractals in PostScript VOORJAAR 2013 65

10 REM Star %!PS-Adobe-3.0 EPSF-3.0

10 P=5 : REM***order*** %%Title: Starfractal by H.A Lauwerier
20 V=4: A=.8*3.141593 : R=.35 %%Transcriptor: Kees van der Laan, kisa1@xs4all.nl, May 2011
30 PSET (0,0) : X=0 : Y=0 /starfractal%reduction angle p (order) v ==> starfractal
40 FOR N=0 TO (V+1)+V^(P-1)-1 {starfractaldict begin
50 M=N : B= N*A : F=0 /v exch cvi def /p exch def /a exch def /r exch def
60 IF M MOD V <> 0 OR F>=P-1 THEN GOTO 80 0 0 moveto
70 F=F+1 : M=M\V : GOTO 60 0 1 v 1 add v p 1 sub exp mul 1 sub
80 X=X+R^(P-F-1)*COS(B) {/n exch cvi def
90 Y=Y+R^(P-F-1)*SIN(B) /m n def /f 0 def
100 LINE -(X,Y) {m v mod 0 ne f p 1 sub ge or
110 NEXT N {exit}
120 END {/f f 1 add def /m m v idiv def}
ifelse
}loop
r p f sub 1 sub exp s 0 rlineto
a rotate
}for %n
end} bind def
/starfractaldict 8 dict def
%
Remarks starfractaldict begin /s {300 mul} def end % scaling
scaling factor has been added in PS .3 144 3 4 starfractal fill % r=.3 a=144 order=3 vertiges-1=4
showpage
user space is rotated by the angle after each line %%EOF
parameter driven

The algorithm is based on that consecutive

line pieces make a constant angle, only the 29
31 28
30 49
51 48
50
line size varies. For the order 5 we have 5 16 18
19
21 23
20
56 58
59
61 63
60

diﬀerent lengths of the line pieces 17

26
22 57 62
53

1 n=0, 256, 512, 768, 1024, … 24

27
25 54
52
55

r n=64, 128, 192, 320, 384, 448, …

r2 n=16, 32, 48, 80, 96, 112, …
71 68 11 8
69 70 9 10
76 78 1 3
r3 n=4, 8, 12, 20, 24, 28, 36, 40, 44, 52, … 79 0
77 2
r4 n=1, 2, 3, 5, 6, 7, 9, 10, 11, 13, … 66 73 6 13

In order to follow the way the drawing has

64 65 74 75 4 5 14 15
67 72 7 12
36 38 41 43
been made sequential numbers have been 39 40
37 42
added in the accompanying illustration. 46 33
44 45 34 35
47 32

Another 5-star composition has been published in my Tiling in PS and Metafont

in MAPS 97.2. I copied the program from the article, adapted it to EPSF, et voilà.
In the middle a composition borrowed from Helmstedt created by a Lindenmayer
production rule in Mathematica. At right a nice illustration from Lauwerier(1990),
which reminds me of Escher’s limit cycles.
66 MAPS 44 Kees van der Laan

Game of Life
Lauwerier(1990) mentions a fractal which he obtained from the Pickover variant of
the Game of Life, made popular by Martin Gardner in a Scientiﬁc American in 1970.
The game is played on a grid. Each node can be alive or dead. Once alive it stays alive.
If dead it comes to life if only one neighbour, N, E, S or W is alive. On each heartbeat
the whole grid is inspected in parallel. Lauwerier’s BASIC program is given below.
***naam: PICK1***
40 DEFINT I, J, K, N, T, X, Y
70 IF SCR=9 THEN XM=320 : YM=175
80 IF SCR=12 THEN XM=320 : YM=240
100 INPUT "NUMBER OF ROWS=", N
120 DIM X(N,N), Y(N,N)
130 X(0,0)=1
140 FOR K=1 TO N-1
150 FOR I=0 TO K : FOR J=0 TO K-I
160 IF X(I,J)=0 THEN GOSUB 220 ELSE GOSUB 270
170 NEXT J : NEXT I
180 FOR I=0 TO K : FOR J=0 TO K-I
190 X(I,J)=Y(I,J)
200 NEXT J : NEXT I : NEXT K
210 A$=INPUT$(1) :END
220 IF I>=1 AND J>=1 THEN T=X(I+1,J)+X(I-1,J)+
X(I,J+1)+X(I,J-1)
230 IF I=0 AND J>=1 THEN T=2*X(1,J)+
X(0,J+1)+X(0,J-1)
240 IF I>=1 AND J=0 THEN T=2*X(I,1)+
X(X+1,0)+X(I-1,0)
250 IF T=1 THEN Y(I,J)=1
260 RETURN
270 PSET (XM+2*I, YTM-2*J),14 : PSET (XM-2*I, YM-2*J),14
280 PSET (XM+2*I, YTM+2*J),14 : PSET (XM-2*I, YM+-2*J),14
290 RETURN : END

. .... . . .
...... .............
.
If we start with 1 alive node then the generations 1, 2, 3, 4 look as follows

Lauwerier’s program is computational intensive.

My translated version could not reproduce Lauwerier’s result in reasonable time.
I simpliﬁed the program by inspecting on each heartbeat only the new contra
diagonals (i + j =constant) in the ﬁrst quadrant.
The 1 …6 generations look
.........
. . . . . .
..................................
. .
. ... ...... .............
. . . . ................
. . . .
............
.
....
Classical Math Fractals in PostScript VOORJAAR 2013 67

%!PS-Adobe-3.0 EPSF-3.0
%%Title: Growth Cell model a la Pickover, simplified
%%Author: Kees van der Laan
...
%%Date: March 2012
%%BoundingBox: -195 -195 195 195 .. . .............................
... .... ...
%%BeginSetup
. .. ...........................................................
%%EndSetup
...... ......... .....
%%BeginPrologue
.. .
. .................. ......................................... ......................
%%DocumentFonts: Times-Roman .. .. .. .. .. .. .. ... .. .. .. .. .. .. ..
/Times-Roman 20 selectfont
... ......................................................................................
/printdot{3 i mul -3 j mul moveto (.) show .......... ... ..... ...
........ ....................... ...........
............
3 i mul +3 j mul moveto (.) show . ......... .............. . .. ......... .........
......... . ...... ......................................... ................
-3 i mul -3 j mul moveto (.) show
-3 i mul +3 j mul moveto (.) show . . .
....... .......... .......... .................. .......... .............. .............
. . . . . .
.
. . . .. . . . . . . . . . . . . . . . . . . . . . . . . .. . . .
}def ...... ..... ...... ...... ...... ...... ...... .......... ...... ...... ...... ...... ...... ..... ......
/alive?{%check whether cell has become alive .........................................................................................................................................................................................................................................................................................
i 1 ge j 1 ge and ...... ...... ...... ...... ...... ...... ...... .......... ...... ...... ...... ...... ...... ...... ......
........... ........... ........... ..................... ............ ........... ...........
{ax i 1 sub n mul j add get ...... .... ...... ....... ... .... ....
ax i n mul j 1 sub add get add}if .......................... .................................................... ............................
i 0 eq j 1 ge and ...... ...... ...... .......... ...... ..................
{2 ax n j add get mul
........... ..................... .........
...... ..........
ax j 1 sub get add}if ................................................................................................
i 1 ge j 0 eq and{ax i 1 sub n mul get}if ...... ...... ...... .......... ...... ...... ......
1 eq{ax i n mul j add 1 put printdot}if ........... ..................... ...........
...... ....... ....
}def%alive? ...................................................
/pickover{% stack integer>=0 the order==> cell pattern ...... .......... ......
/n exch def .....................
/ax n 1 add dup mul array def %array
..........
1 1 n n mul{/k exch def ax k 0 put}for ax 0 1 put%initialize
/i 0 def /j 0 def printdot
1 1 n 1 sub {/k exch def
0 1 k{/i exch def /j k i sub def%contradiagonal
alive?
}for%i
}for%k
}def
%%EndProlog
%
% Program
%
64 pickover showpage
%%EOF

The point Lauwerier wanted to make — the game yields fractal patterns — is also
obtained by this simpliﬁed game.

Annotated References
- An introductory survey: http://en.wikipedia.org/wiki/Dragon_curve.
- Adobe Red, Green and Blue Books. The musts for PS programmers.
- Biography of H.A. Lauwerier: http://bwnw.cwi-incubator.nl/cgi-bin/uncgi/alf.
- Gleisk, J(1987): CHAOS — making a new science. Penguin.
(An introduction to and survey of the world of non-linearity, strange attractors
and fractals.)
- Goossens, M(2007, sec ed) et. al.: LATEX Graphics Companion. ISBN 978 0 321
50892 8.
- Helmstedt, J(2011): A New Method of Constructing Fractals and Other Graphics.
The Mathematica Journal. (Nice examples of Lindenmayer systems, for which
Lauwerier’s KRONKEL can be used.)
68 MAPS 44 Kees van der Laan

- Jackowski, B, P. Strelczyk, P. Pianowski(1995-2008): PSView5.12. WWW.

bop@bop.com.pl. (Extremely fast previewer for .eps among others, which allows
PSlib(rary) inclusion via the run command).
- Knuth, D.E, T. Larrabee, P.M. Roberts(1989): Mathematical Writing. MAA notes
14. The Mathematical Association of America.
- Knuth, D.E(1990, 10th printing): The TEXbook. Addison-Wesley. ISBN
0-201-13447-0. (A must for plain TEXies.)
- Lauwerier, H.A(1987): FRACTALS — meetkundige figuren in eindeloze her-
haling. Aramith. (Contains programs in BASIC. Lauwerier, H.A (1991): Frac-
tals: Endlessly Repeated Geometrical Figures, Translated by Sophia Gill-Hoff-
stadt, Princeton University Press, Princeton NJ1. ISBN 0-691-08551-X, cloth.
ISBN 0-691-02445-6 paperback. "This book has been written for a wide audi-
ence … " Includes sample BASIC programs in an appendix. Audience: Instruc-
tors, (high-school) students, and the educated layman.)
- Lauwerier, H.A(1988): The Pythagoras Tree as Julia Set. CWI-Newsletter.
- Lauwerier, H.A(1989): Oneindigheid — een onbereikbaar ideaal. Aramith. ISBN
90 6834 055 7. (Audience: Instructors, (high-school) students, and the educated
layman.)
- Lauwerier, H.A(1990): Een wereld van FRACTALS. Aramith. ISBN 90 6834
076 X. (Sequel and updated version of Lauwerier(1987). Audience: Instructors,
(high-school) students, and the educated layman.)
- Lauwerier, H.A(1994): Spelen met Graphics and Fractals. Academic Service.
ISBN 90 395 0092 4. (An inspiring book with Math at the high school level for
a wide audience; the BASIC programs I consider outdated for direct use. Audi-
ence: Instructors, (high-school) students, and the educated layman.)
- Peitgen, H.O, H.Jürgens, D. Saupe(2004 sec. ed.): Chaos and Fractals. New fron-
tiers of Science. (Images of the fourteen chapters of this book cover the central
ideas and concepts of chaos and fractals as well as many related topics includ-
ing: the Mandelbrot set, Julia sets, cellular automata, L-systems, percolation and
strange attractors. This new edition has been thoroughly revised throughout.
The appendices of the original edition were taken out since more recent publi-
cations cover this material in more depth. Instead of the focused computer pro-
grams in BASIC, the authors provide 10 interactive JAVA-applets for this second
edition via http://www.cevis.uni-bremen.de/fractals. An encyclopedic work.
Audience: Accessible without mathematical sophistication and portrays the new
fields: Chaos and fractals, in an authentic manner.)
- Swanson, E(1986, revised ed): Mathematics into Type. American Mathematical
Society.
- Szabó, P(2009): PDF output size of TEX documents. Proceedings Eu-
roTEX2009/ConTEXt, p57–74. (Various tools have been compared for the pur-
pose.)
- Van der Laan, C.G(1992): LIFO and FIFO sing the Blues. MAPS 92.2.
- Van der Laan, C.G(1995): Publishing with TEX. Public Domain. (See TEX
archives. BLUe.tex comes with pic.dat the database of my pictures in
TEX-alone.)
- Van der Laan, C.G(1997): Tiling in PostScript and MetaFont — Escher’s wink.
MAPS 97.2.
- Van der Laan, C.G(unpublished, BachoTEX workshop): TEXing Paradigms. (A
plea is made for standardized macro writing in TEX to enhance readability and
correctness.)
- Van der Laan, C.G(submitted EuroTEX2012): Julia fractals in PostScript.
- Veith, U(2009): Experiences typesetting mathematical physics. Proceedings Eu-
roTEX2009/ConTEXt, p31–43. (Practical examples where we need to adjust TEX’s
automatic typesetting.)
Classical Math Fractals in PostScript VOORJAAR 2013 69

Conclusions
It was pleasure, educative and inspiring to read Lauwerier’s booklets. Some of his
algorithms have found a wider audience by converting his BASIC codes into Post-
Script, hopefully.
I don’t know how to include the results of the BASIC programs elegantly in pub-
lications. The results of the PS programs can be easily included in pdf(La)TeX, Word,
… documents.
PS’ variable user space and recursion alleviated programming, with concise defs
and programs as readable as literature, but … be aware of its subtleness. PostScript’s
variable user space was the key to my adaptation of production rules. Because of PS’
subtleness not many people program in PS, I presume, or … do they consider it of
too low-level?
In programming self-similarity the awareness of orientation is paramount. I did
not find classical Math fractals in PS on the WWW, only one Sierpiński curve in Java.
Lauwerier’s analysis — associating binary, ternary, … tree structures with binary,
ternary, … numbers, is an eye-opener. In his, and my, programs all the self-similar
sub-curves are draw anew. In Metafont/-Post we could just build the paths and splice
them suitably into paths of higher order, as I did in the past with the Pythagoras Tree
in Metafont.
‘Het Wiskunde boek’ states that fractals have renewed and raised interest in Math-
ematics.
Before publishing consult the Wikipedia on aspects of the subject as well as Wol-
fram’s knowledge base http://www.wolframalpha.com.
TEX mark up For the symbols of the number systems I, N, Q, R, C, which curiously
are not provided for in plain TEX, I use the the AMS (blackboard) font msbm10.
L R
The = and = composed relational operators are marked up by \mathrel {\mathop
=^{\rm L}} and not by $\buildrel\rm L \over=$, TEXbook p437; the latter is OK for
the stacked composed symbol as such.
For typesetting tables \halign and the tabbing mechanism have been used
(TEXbook ch22). The 11-element of one of the tables needs an oblique line. I provided
for this in PS, which is simpler and not restricted by obliqueness. In 1995, in my PWT
guide, I used the GKP macros for this, which suffer from the same inconvenience as
LATEX’s picture environment: restricted obliqueness.
A blank line before display math yields too much white space! This blank line is
important, though, in order to avoid widows.
Locally I have used for parallel listings of program texts vbox-s next to each other,
which inhibits proper page breaks. I don’t know how to provide macros for lo-
cal elegantly marked up multi-column texts, which allow page breaks. (I also tried
\valign, alas in vain.) My inserted pictures suffer from the same inconvenience as in
Word: changing the text might disturb the layout, such that the pictures will become
ill-placed.
As known, I could not use footnotes from within a vbox; kludged around.
In TEXworks I used the Terminal font in the edit window with the pleasing effect
that comments remain vertically aligned in the .pdf window.
Conversion of my TEX script into Word made me (hands-on) aware of the differences
between TEX and Word. If you are after utmost accurate, user-controlled typeset
Mathematics then TEX is to be preferred. For bread and butter Mathematics Word
can do, especially with Cambria, I presume. I did not find in Word (MS equation
3.0) the possibility to discern between displayed Math and in-line Math. Tables in
Word are tricky, the WYSIWYG approach does not always yield the table layout you
are after. As in TEX I can’t make an appropriate 11-element. I could not handle the
inclusion of a PS made 11-element in Word. Program texts, as columns in a table,
70 MAPS 44 Kees van der Laan

don’t suffer from difficulties in allowing page breaks. A pre-index, as is usual with
hyper-geometric functions, I could not nicely typeset with MS equation 3.0.
Inclusion of the .jpg figures and .pdf objects went smoothly. I had to convert .png
objects. The inclusion of EPSF object option did not work on my PC, though the
option is available. It invoked Adobe Illustrator CS2 12.0.0 and fell silent. The same
EPSF invoked by AI directly worked. Maybe incompatible versions? Neglecting su-
perfluous spaces, which TEX does automatically, has been lost in the conversion. A
local change in Word might change the document more than local, beyond user con-
trol. I don’t know how to switch off, or change, pre-settings, such as: don’t underline
automatically WWW addresses, maybe by de-activating the option WWW addresses
as hyperlinks?
Conversion also entailed splitting up the original (concept) paper and rewriting
the parts into 2 new papers. Converting back into TEX, after changes were made, was
more difficult than converting into Word.
After I had finished I became aware of Acrobat Pro X which also converts .pdf
into a Word document.

Acknowledgements
Thank you Adobe for your maintained, adapted to LanguageLevel 3 since 1997, good
old, industrial standard PS and Acrobat Pro (actually DISTILLER) to view it, Don
Knuth for your stable plain TEX, Jonathan Kew for the TEXworks IDE, Hàn Thế Thành
for pdf(La)TEX, Hans Lauwerier for your nice booklets with so many inspiring ex-
amples of fractals.
Thank you Jos Willink for proofing, Wim Wilhelm for drawing my attention to
cellulaire automata, and the LATEX graphics environment asymptote which he in-
tegrated in TEXnicCenter; I don’t have experience with them, but they may be of
interest for LATEX users. MAPS editors for improving my use of English and Taco
Hoekwater for procrusting my plain TEX note into MAPS format.
Thank you Zinadia Nikolaevna Gulka for inviting me to submit a paper or two for
the ‘Informatsionnie Texnologii i Matematicheskoe Modelirovanie’ journal’, and her
co-worker for stimulating me to convert the TEX marked up ASCII source of an early
version of this note into the required Word. The invitation stimulated me to adapt
and revise the material. Thank you GUST for publishing a previous version of this
note in the BachoTEX2012 proceedings. Thank you Svetlana Morozova for prompting
me in the use of Word.
IDE My PC runs 32 bits Vista, with Intel Quad CPU Q8300 2.5GHz assisted by
8GB RAM. I visualize PS with Acrobat Pro 7. My PS editor is just Windows ‘klad-
blok (notepad).’ I use the EPSF-feature to crop pictures to their BoundingBox, ready
for inclusion in documents. For document production I use TEXworks IDE with the
plain TEX engine, pdfTEX, with as few as possible structuring macros taken from
my BLUe.tex — adhering minimal TEX markup. I use the Terminal font in the edit
window with the pleasing effect that comments remain vertically aligned in the .pdf
window.
For checking the spelling I use the public domain en_GB dictionary and hyphen-
ation patterns en_GB.aff in TEXworks.
Prior to sending my PDF’s by email the files are optimized towards size by Acrobat
Pro.
The bad news with respect to .eps into .pdf conversion is that the newest Acrobat
10 Pro X does not allow for the run command for library inclusion.
Classical Math Fractals in PostScript VOORJAAR 2013 71

Notes
1. Alas, the \psfig has been lost in pdfTEX. Happily, ConTEXt and LuaTEX allow
direct EPSF inclusion.
2. Lauwerier(1989) narrates what mathematicians thought about the ∞-concept
through the ages from the ancient Greeks onward.
3. Acrobat Pro X does not allow for the library inclusion via run, alas :-(. BASIC is
interactive, PS is batch-oriented.
4. Barnsley is famous for his fern fractal. In his later works he is more ambitious
and constructs fractals given a picture, on demand.
5. The ‘ﬁxed-point’ of the production rule is the fractal. At right my old TEX code is
displayed with its result.
6. The ‘ﬁxed-point’ of the production rule is the real fractal.

My case rests, have fun and all the best.

Kees van der Laan
Hunzeweg 57, 9893PB Garnwerd, Gr, NL
kisa1@xs4all.nl
72 MAPS 44 Kees van der Laan

Appendix: Fractal Dimension

The need arose to associate various fractal curves with numbers, to characterize them.
Mathematicians came up with definitions which were generalizations and compatible
extensions of the classical, topological dimension notion. The fractal dimension à la
Kolmogorov(1958) is based upon covering the fractal with a grid and counting the cells
of the grid which contain points of the fractal, the so-called box-counting dimension.
The definition reads
D = lim logN⁄log(1⁄s),
s→0
with N the number of cells which contain points of the fractal and s the size of the
side of a cell. Let us calculate the (fractal) dimension of a square in order to verify
the compatibility of the definition with the classical fact. Cover the (unit)square with a
grid which has s as side of a cell, then we have 1⁄s2 cells which cover the unit square,
and therefore D = log(1⁄s2 )⁄log(1⁄s) = 2, QED.
All plane-filling fractal curves have fractal dimension D = 2.
For the calculation of the fractal dimension of fractals defined by contracted map-
pings, with contraction factors f1 , f2 , f3 , … Lauwerier(1990) mentions the Hausdorff
formula
D
∑ fi = 1.
i
For the Lévy fractal, which can be defined by 2 contractions each with contraction
factor 1⁄√2, we arrive at
2 ∗ (1⁄√2)D = 1 → D = 2.
For the von Koch fractal, which can be defined by 4 contractions each with contraction
factor 1⁄3, we arrive at
4 ∗ (1⁄3)D = 1 → D = log4⁄log3 ≈ 1.26.
For the Sierpiski sieve, which can be defined by 3 contractions with contraction factors
1⁄2, we arrive at
3 ∗ (1⁄2)D = 1 → D = log3⁄log2 ≈ 1.58.
For the Menger sponge, which can be defined by 8 contractions each with contraction
factor 1⁄3, we arrive at
8 ∗ (1⁄3)D = 1 → D = log8⁄log3 ≈ 1.89.
The details of the definition of the fractal dimension D by Hausdorff (1919) are
cumbersome, and serve a theoretical need. The calculation of other fractal dimensions:
self-similarity dimension, and compass dimension (for coast lines) goes beyond the
scope of this paper, see Peitgen c.s.(2004). My purpose is that one knows about the
concept as a grey box with nodding knowledge, that one knows some fractal dimensions
for simple cases, and … that one does not become frightened or confused on encounter.
For a survey of various fractals and their dimensions see Peitgen c.s.(2004) and/or
http://en.wikipedia.org/wiki/List_of_fractals_by_Haussdorf_dimension.
Finally, do you know that the path of Brownian movement is a fractal with fractal
dimension 2, and … do you know that the Cantor Dust has Hausdorf dimension DH =
log2/log3 ≈ .63, which differs from the fractal dimension D ≈ .69?

Appendix: Cantor Dusts

The Cantor Dust has been shown at the beginning of this note. The idea is that each
line is replaced by its left and right third parts. This pattern can be repeated yielding
what is called a fractal nowadays. (It is included in pic.dat, the library of (TEX-alone)
pictures which come with BLUe.tex.)
In Publishing with TEX(1995) the Cantor Dust example by TEX alone was included.
Classical Math Fractals in PostScript VOORJAAR 2013 73

The PS code for Cantor Dust of order n has a similar production rule as for the von
Koch fractal:
CDn = [S 1 1 CDn−1 ] ⊕ [S 1 1 T 2s 0 CDn−1 ] with
3 3 3 3 3
CD0 the initial line,
CDn the Cantor Dust of order n,
⊕ splice operator, meaning add properly the second piece to the set,
[ open a new GS on the GS stack,
] remove current graphics state from the GS stack and recall previous,
Sa,b means scale US by a and b, in x- and y-direction
Ta,b means translate US by a and b, in x- and y-direction.
The above PS production rule transcribes systematically into the following PS def.5
%!PS-Adobe-3.0 EPSF-3.0
/cd%integer n>=0 ==> Cantor Dust of order n
{1 sub dup -1 eq
{0 0 moveto s 0 lineto stroke}
{gsave .3333 1 scale dup cd grestore
gsave .3333 1 scale 2 s mul 0 translate dup cd grestore
}ifelse 1 add
}def
%%EndProlog
%
% Program
%
%...
/s 300 def 4 cd pop
305 -2 moveto TR10 setfont (Cantor) show
0 -3 rmoveto TR7 setfont (4) show 0 -10 translate
%...
showpage
%%EOF
%Borrowed from BLUe's pic.dat
\newcount\x\newcount\y\newcount\width
\newdimen\unitlength
\def\E#1{\hbox to 0pt{\kern\x\unitlength
\vbox to 0pt{\vss\hrule width#1\unitlength
\kern\y\unitlength
}\hss
}\advance\x#1 }
\def\cf{\ifnum0=\width \fc\fi
{\E{\the\width}}\divide\width3
\advance\y-3
{\cf}\advance\x\width
\advance\x\width
{\cf}\relax}%
\def\fc#1\relax{\fi}%
$$\width243 \unitlength1pt
\x-\width \divide\x2
\cf$$

The ‘ﬁxed point’ of the production rule is the Cantor Dust fractal. For just showing a
few approximations of the Cantor Dust the TEX code will do. Lauwerier(1987) provides
a tiny BASIC program KAM. He also associates the Cantor dust with the trinary number
system because the interval is divided in 3 pieces, repeatedly. The Cantor dust of [0, 1]
consists of the trinary fractions where only the digits 0 and 2 occur, Lauwerier(1987,
p26). An eye-opener!
Cantor Dust as IFS Lauwerier(1989, ch8) constructs the Cantor Dust of high order by
the IFS
74 MAPS 44 Kees van der Laan

and
L R
xn+1 = xn ⁄3 xn+1 = xn ⁄3 + 2⁄3 n = 0, 1, 2, … x0 = 1⁄3.

Again an eye-opener! His tiny program and my conversion read

X=.3 'start
FOR K=1 TO 100
IF RND<.5 THEN X=X/3 ELSE X=(X+2)/3
PSET(X, 0)'scale X if wanted
NEXT K
END
%!PS-Adobe-3.0 EPSF-3.0
%%Title: Cantor Dust via IFS
%%Author: H.A. Lauwerier(1989): Oneindigheid
%%Transcriptor: C.G. van der Laan, may 2012
%%BoundingBox: -1 39 1010 70
%%BeginSetup
%%EndSetup
%
% Program
%
/Courier 10 selectfont
22121943 srand /x .3 def%start
100{rand 1073741823 gt{/x x 2 add 3 div def}
{/x x 3 div def}ifelse
x 1000 mul 50 moveto (.) show
}repeat
showpage
%%EOF

If you, kind reader, shrug shoulders about paying so much attention to such a tiny
problem, I can only say that if you don’t analyse tiny problems deeply, your solutions
of bigger problems will lack ingenuity.
Generalization to 2D yield the so-called Sierpiński carpets.
The algorithm reads: divide a square in 32 equal sub-squares and delete the middle,
or the middle cross, and do this repeatedly for the remaining 8 squares.
At the EuroTEX95 Bogusłav Jackowski showed his variant, probably in connection
with his mftoeps program, which transforms Metafont code into PS. (I did not use
mftoeps, because it was PC-biased, and I had a Mac Powerbook 150.) This was at the
time that MetaPost, the preprocessor for PS with Metafont-biased user-language, was
not yet released in the public domain.
In my opinion he was on the right track: PS is mandatory for graphics to be included
in documents. It is just a pity he did not pursue that PS alone, or better EPSF, is
suitable for constructing graphics to be included in documents. His collaborators, Pjotr
and Pjotr, pursued PS in PSview.
I programmed the Sierpiński carpet, it is not the gasket, in TEX and in Metafont
after the conference. For historical reasons I have included the programs below. The
black-and-white ﬁgure is created by TEX-alone on-the-ﬂy. (I just copied it from the
revised 1996 FIFO-article, and see,… it still works. The MF program also still works on
my museum Mac Powerbook 150 of 1995!)
The picture at right has been borrowed from the WWW. It illustrates a generalization
of the good old Cantor Dust into 2.5D, also called Menger sponge. Both pictures are
more interesting and more beautiful than the original 1D Cantor Dust.
Classical Math Fractals in PostScript VOORJAAR 2013 75

tracingstats:=1;proofing:=1;screenstrokes; %
pickup pencircle scaled 1;
def sierpinskisquare (expr s, p)=
if s>5:unfill unitsquare scaled .333s
shifted (p+.333s*(1,1));
sierpinskisquare(.333s, p);
sierpinskisquare(.333s, p+(.333s,0));
sierpinskisquare(.333s, p+(.667s,0));
sierpinskisquare(.333s, p+(0,.333s));
sierpinskisquare(.333s, p+(0,.6673s));
sierpinskisquare(.333s, p+(.667s,.333s));
sierpinskisquare(.333s, p+(.667s,.667s));
sierpinskisquare(.333s, p+(.333s,.667s));
fi enddef;
%
s=100; fill unitsquare scaled s;
sierpinskisquare(s,origin);
showit;
end
\newdimen\x\newdimen\y\newdimen\size\newdimen\lsize
\def\sier{\ifdim\size<35pt \reis\fi
\divide\size3
{\sier}{\advance\x\size\sier}{\advance\x2\size\sier}%
{\advance\y\size{\sier}{\advance\x2\size\sier}}%
\advance\y2\size{\sier}{\advance\x\size\sier}%
\advance\x2\size\sier}
\def\reis#1\sier{\fi\putatxy\draw}
%with auxiliaries
\def\putatxy#1{\vbox to0pt{\vss
\hbox to0pt{\kern\x#1\hss}\kern\y}}
\def\draw{{\lsize\size\divide\lsize3
\rlap{\vrule height\size width\lsize
\vbox to\size{\hrule width\lsize height\lsize\vss
\hrule width\lsize height\lsize}%
\vrule height\size width\lsize}}}
$$\vbox to120pt{\vss
\offinterlineskip\size120pt\x=-.5\size\y0pt
\sier}$$

A PS program for a 2.5D Cantor Dust is cumbersome, because it has to deal with
projection and has to handle hidden lines.

Appendix: Hilbert Curve

Hilbert curves H1 … H6 have been shown in the introduction. Wirth(1975) I consider
a starting point for programming a Hilbert curve. Wirth could not know about the
rotation of US facility, nor did PS exist. The self-similarity property of the H-curve,
i.e. a curve is composed of (rotated) curves of one order lower, he programmed by
four rotated instances in (recursive) procedures A, B, C and D, which entailed a more
complex recursion scheme. For those who don’t own the 1975 book I have included
the program and the procedure A. (Procedures B, C and D are similar.)
76 MAPS 44 Kees van der Laan

The left PS-program below is based on the production rule6

Hi = m H90
i−1 ⊕ ↑ ⊕ Hi−1 ⊕ → ⊕ Hi−1 ⊕ ↓ ⊕ m Hi−1 ,
−90 for 1 = 1, 2, …
where ⊕ means spliced, the superscript denotes the rotation angle, the arrows ↑ → ↓
mean draw a segment in the direction North, East, and South. In order to splice
correctly, the rotated copies have also to be mirrored, which is indicated by the pre-index
m.

%!PS-Adobe-3.0 EPSF-3.0 program Hilbert(pf,output);

%%Title: Hibert curve, Feb 2012 {plot Hilbert curves of orders 1 to n. Wirth(1975).}
%%Author: Kees van der Laan, kisa1@xs4all.nl const n= 4; h0=512;
%%BoundingBox: -1 -1 321 151 var i,h,x,y,x0,y0: integer;
%%BeginSetup pf: file of integer; {plotfile}
%%EndSetup procedure A(i: integer);
%%BeginProlog begin if i>0 then
/-s s neg def begin D(i-1); x:=x-h; plot;
/lineN{0 0 moveto 0 s lineto currentpoint stroke translate}def A(i-1); y:=y-h; plot;
/lineS{0 0 moveto 0 -s lineto currentpoint stroke translate}def A(i-1); x:=x+h; plot;
/lineE{0 0 moveto s 0 lineto currentpoint stroke translate}def B(i-1);
% end
/Hilbert{%on stack order >=1 ==> Hilbert curve end;
%s size of line segment (global), {B, C and D similar for rotated instances}
1 sub dup 0 gt begin startplot;
{90 rotate 1 -1 scale Hilbert 1 -1 scale -90 rotate i:=0; h:= h0; x0:=hdiv2; y0:=x0;
lineN Hilbert repeat {plot Hilbert curce of order i}
lineE Hilbert i:=i+1; h:= h div 2;
lineS x:=x0 + h div 2;
-90 rotate 1 -1 scale Hilbert 1 -1 scale 90 rotate y:=y0 + h div 2;
}if 1 add%reset order setplot;
} def A(i)
%%EndProlog until i=n 6 7 10 11

% endplot

21
%Program ---the script--- end.

9
%

41
2 3 4 13

3
/s 10 def %size of segment
1 Hilbert pop
4

51
1 1 16
1

2
s 0 translate 2 Hilbert pop
2 s mul 0 translate 3 Hilbert pop
3 s mul 0 translate 4 Hilbert pop
showpage
%%EOF

The US facility of PS facilitates programming of these kinds of curves. Moreover, we

don’t need a plotfile in PS. After executing the PASCAL code we still have to create
a .pdf file suitable for inclusion in documents.
Lauwerier(1990) provides a BASIC program, Peanol, for the Hilbert curves, re-
stricted to orders ≤ 5, via the Turtle Graphics method, i.e. drawing forward, meaning
the broken line is obtained by drawing the segments in increasing order 1, 2, 3, … . The
above program draws also forward by backtracking, a short of LIFO, Last-In-First-Out.
In Peanol the direction numbers are stored in an array of size 3411. Large enough for
practical purposes. (The program above does not need the explicit direction numbers.)
The Peano curve, as shown at the beginning of this note, can be programmed along
the same lines as for the Hilbert broken line above or via the Turtle Graphics casu quo
Lindenmayer approach à la Lauwerier, which I leave as an exercise to the reader.
Application of plane-filling curves occurs in discretization of images where instead of
a Cartesian grid the plane filling curve is used.
Classical Math Fractals in PostScript VOORJAAR 2013 77

Hilbert curve in Metafont and Joseph Romanovski’s (1995) PS code

%cgl, 1995 ...

tracingstats:=1; proofing:=1;screenstrokes;autorounding:=0; /S{0 R rlineto currentpoint stroke moveto}def
pickup pencircle scaled 0.2pt; /T{90 rotate}def
def openit = openwindow currentwindow /TM{T 1 -1 scale}def
from origin to (screen_rows,screen_cols) at (-40s,15s)enddef; /H{TM dup 0 gt
path p; s=10; {1 sub
sz=0;p:=origin;%H_0 size and path H S
n=5; %Order of H-curve TM H S
for k=1 upto n:%H_1,...H_n consecutively H T S
p:= p transformed (identity rotated 90 -1 1 scale H 180 rotate
reflectedabout (origin,up))-- 1 add
p shifted ((-sz-1)*s,0)-- }if TM
p shifted ((-sz-1)*s,(-sz-1)*s)-- }def
p transformed (identity rotated -90 /R 8 def 100 100 moveto 6 H pop
reflectedabout (origin,up) shifted (-sz*s,(-2sz-1)*s)); showpage
sz:=2sz+1;
clearit;draw p; showit;
endfor
end

Joseph’s code is a bit cryptic (but is similar to mine). He uses short names, which does
not make sense, makes the code diﬃcult to read, which is considered a bad practice.
But, … he was on the right track, also with the use of colours via PS.
Conclusion. Such a simple(?) problem yielded already a few variants, of which most
suﬀer from bad readability, because they are not biased by a production rule, IMHO.

Appendix: Sierpiński islands

Sierpiński islands S1 … S3 have been shown in the introduction, in overlay.
Wirth(1975), I consider a starting point for programming a Sierpiński island. Wirth
did not have the rotation of US facility, nor did PS exist.
Algorithm: The curve is closed and is composed of 4 (90∘ rotated) broken lines
connected by lines in the direction SE, SW, NW and NE. The program consists of
2 parts: ﬁrst the generation of a side and second the appropriate splicing of rotated
copies.
The PS program is based on the following production rule for a side
Si = Si−1 ⊕ SE ⊕ S−90 90
i−1 ⊕ E ⊕ Si−1 ⊕ NE ⊕ Si−1 , for i = 1, 2, …
where ⊕ means spliced, the superscript denotes the rotation angle. and SE, E, NE mean
draw a line in the direction South-East, East, and North-East.

%!PS-Adobe-3.0 EPSF-3.0
%%Title: Sierpinski island Side, March2012
%%Author: Kees van der Laan
%% Affiliation: kisa1@xs4all.nl
%%BoundingBox:-1 -82 287 2
%%BeginSetup
%%EndSetup
%%BeginProlog
/s' s 1.4142 div def
/NE{0 0 moveto s' dup lineto currentpoint stroke translate}def
/SE{0 0 moveto s' dup neg lineto currentpoint stroke translate}def
/E {0 0 moveto s 0 lineto currentpoint stroke translate}def
/SideS{%on stack order >=1 ==> Sierpinski side
%s size of line segment (global)
1 sub dup 0 ge
{ dup SideS
SE
dup -90 rotate SideS 90 rotate
E
78 MAPS 44 Kees van der Laan

dup 90 rotate SideS -90 rotate

NE
dup SideS
}if 1 add} def
0 setlinejoin 1.415 setmiterlimit 1 setlinecap
%%EndProlog
%
%Program ---the script---
%
/s 10 def % size of segment
1 SideS pop
s 0 translate 2 SideS pop
s 0 translate 3 SideS pop
showpage

Wirth programmed the four rotated instances in (recursive) procedures A, B, C and D,

which entailed a more complex recursion scheme.

Sierpiński islands 0 … 3
For those who don’t own the 1975 book I have included Wirth’s program, translated
into Metafont in 1995.

%!PS-Adobe-3.0 EPSF-3.0 %Translation of Wirth's code into Metafont

%%Title: Sierpinski island, March 2012 tracingstats:=1;proofing:=1;screenstrokes;
%%Author: Kees van der Laan pickup pencircle scaled 0.01pt;
%%Affiliation: kisa1@xs4all.nl s=10; path p;
%%BoundingBox:-10 -185 307 2 def openit = openwindow currentwindow from origin
%%BeginSetup to (screen_rows,screen_cols) at (-5s,5s)enddef;
%%EndSetup def A expr k=if k>0:
%%BeginProlog A(k-1);draw z--hide(x:=x+h;y:=y-h)z;
(C:\\PSlib\\PSlib.eps) run %loads /SideS and auxiliaries from PSlib B(k-1);draw z--hide(x:=x+2h)z;
%%EndProlog D(k-1);draw z--hide(x:=x+h;y:=y+h)z;
% A(k-1); fi enddef;
%Program ---the script--- def B expr k=if k>0:
% B(k-1);draw z--hide(x:=x-h;y:=y-h)z;
/s 20 def % size of segment C(k-1);draw z--hide(y:=y-2h)z;
4{ SE -90 rotate}repeat%S_0 A(k-1);draw z--hide(x:=x+h;y:=y-h)z;
1.5 s mul 0 translate /s s 2 div def B(k-1); fi enddef;
4{ 1 SideS pop SE -90 rotate}repeat%S_1 def C expr k=if k>0:
4.5 s mul 0 translate /s s 2 div def C(k-1);draw z--hide(x:=x-h;y:=y+h)z;
4{ 2 SideS pop SE -90 rotate}repeat%S_2 D(k-1);draw z--hide(x:=x-2h)z;
10.5 s mul 0 translate /s s 2 div def B(k-1);draw z--hide(x:=x-h;y:=y-h)z;
4{ 3 SideS pop SE -90 rotate}repeat%S_3 C(k-1); fi enddef;
showpage def D expr k=if k>0:
%%EOF D(k-1);draw z--hide(x:=x+h;y:=y+h)z;
A(k-1);draw z--hide(y:=y+2h)z;
C(k-1);draw z--hide(x:=x-h;y:=y+h)z;
D(k-1); fi enddef;
def sierpinski expr k=%k is order of curve
The S0 …S3 pictures obtained by if k>0: A(k);draw z--hide(x:=x+h;y:=y-h)z;
the given program are shown above. B(k);draw z--hide(x:=x-h;y:=y-h)z;
C(k);draw z--hide(x:=x-h;y:=y+h)z;
D(k);draw z--hide(x:=x+h;y:=y+h)z;
fi enddef;
z=origin; h=16;
sierpinski2; showit; end

Full download First Course in Mathematical Modeling 5th ed. A pdf docx
100% (5)
Full download First Course in Mathematical Modeling 5th ed. A pdf docx
34 pages
Step by Step Binance Guide
0% (1)
Step by Step Binance Guide
96 pages
Plain TeX With Postscript
No ratings yet
Plain TeX With Postscript
22 pages
XDS - Modula 2.IDE - User.guide - en
No ratings yet
XDS - Modula 2.IDE - User.guide - en
61 pages
LLVM+GHV Thesis
No ratings yet
LLVM+GHV Thesis
73 pages
Physics Lab Manual PDF
No ratings yet
Physics Lab Manual PDF
57 pages
Full Download Quantum Computing by Practice Python Programming in The Cloud With Qiskit and IBM Q 2nd Edition Silva PDF
100% (3)
Full Download Quantum Computing by Practice Python Programming in The Cloud With Qiskit and IBM Q 2nd Edition Silva PDF
73 pages
Click Here For Download: (PDF) Mercury's Perihelion From Le Verrier To Einstein (Oxford Science Publications)
No ratings yet
Click Here For Download: (PDF) Mercury's Perihelion From Le Verrier To Einstein (Oxford Science Publications)
2 pages
Department of Mathematics Manipal Institute of Technology, Manipal
No ratings yet
Department of Mathematics Manipal Institute of Technology, Manipal
2 pages
Download Full Riemannian Geometry and Geometric Analysis 7th Edition Jürgen Jost (Auth.) PDF All Chapters
50% (2)
Download Full Riemannian Geometry and Geometric Analysis 7th Edition Jürgen Jost (Auth.) PDF All Chapters
55 pages
An Introduction To Q-Difference Equations
No ratings yet
An Introduction To Q-Difference Equations
141 pages
Of Functions: An Atlas
No ratings yet
Of Functions: An Atlas
708 pages
Free Ebooks Download
No ratings yet
Free Ebooks Download
31 pages
Cvitanovic Et Al. Classical and Quantum Chaos Book (Web Version 9.2.3, 2002) (750s) - PNC
No ratings yet
Cvitanovic Et Al. Classical and Quantum Chaos Book (Web Version 9.2.3, 2002) (750s) - PNC
750 pages
SHURMAN Multivariable Calculus
No ratings yet
SHURMAN Multivariable Calculus
523 pages
Download ebooks file (Ebook) The Little Typer by Daniel P. Friedman; David Thrane Christiansen ISBN 9780262536431, 0262536439 all chapters
100% (3)
Download ebooks file (Ebook) The Little Typer by Daniel P. Friedman; David Thrane Christiansen ISBN 9780262536431, 0262536439 all chapters
71 pages
LLVM Clang - Advancing Compiler Technology
No ratings yet
LLVM Clang - Advancing Compiler Technology
28 pages
Zhanghongbo Ocaml Book
No ratings yet
Zhanghongbo Ocaml Book
422 pages
Org Mode Teaser
No ratings yet
Org Mode Teaser
24 pages
Notes On Adjoint Methods MIT
No ratings yet
Notes On Adjoint Methods MIT
6 pages
Instant Download Understanding Analysis by Stephen Abbott PDF All Chapters
100% (4)
Instant Download Understanding Analysis by Stephen Abbott PDF All Chapters
24 pages
Instant Ebooks Textbook (Ebook PDF) Fundamentals of Complex Analysis: With Applications To Engineering and Science 3rd Edition Download All Chapters
80% (5)
Instant Ebooks Textbook (Ebook PDF) Fundamentals of Complex Analysis: With Applications To Engineering and Science 3rd Edition Download All Chapters
37 pages
Symbolic C
No ratings yet
Symbolic C
683 pages
Orthogonal Polynomials (In Matlab) : Walter Gautschi
No ratings yet
Orthogonal Polynomials (In Matlab) : Walter Gautschi
23 pages
Programming Forth
No ratings yet
Programming Forth
198 pages
Org-Mode Basics: Ben Maughan October 9, 2015
No ratings yet
Org-Mode Basics: Ben Maughan October 9, 2015
8 pages
Designing and Building Parallel Programs
No ratings yet
Designing and Building Parallel Programs
371 pages
L TEX Reference Card: 1 Greek Letters
No ratings yet
L TEX Reference Card: 1 Greek Letters
4 pages
Geometric Algebra For Electrical Engineers
No ratings yet
Geometric Algebra For Electrical Engineers
286 pages
Arutyunov Notes
100% (1)
Arutyunov Notes
142 pages
Julia As A Portable High-Level Language For Numerical Solvers of Power Flow Equations On GPU Architectures
100% (1)
Julia As A Portable High-Level Language For Numerical Solvers of Power Flow Equations On GPU Architectures
10 pages
Webster Theory of Electricity and Magnetism, Being Lectures On Mathematical Physics (1897)
No ratings yet
Webster Theory of Electricity and Magnetism, Being Lectures On Mathematical Physics (1897)
600 pages
sm381_bk01_itm5124952 (2)
No ratings yet
sm381_bk01_itm5124952 (2)
288 pages
GDC09 Abrash Larrabee+Final
No ratings yet
GDC09 Abrash Larrabee+Final
116 pages
Book 12 Quantum Mechanics of Many-Particle Systems: Atoms, Molecules - and More
No ratings yet
Book 12 Quantum Mechanics of Many-Particle Systems: Atoms, Molecules - and More
173 pages
Algebraic Geometry - Dieudonné
100% (1)
Algebraic Geometry - Dieudonné
89 pages
Unix Network Programing in C
100% (3)
Unix Network Programing in C
28 pages
On Rice S Theorem: 1 Properties
No ratings yet
On Rice S Theorem: 1 Properties
7 pages
Basilisk Tutorial
0% (1)
Basilisk Tutorial
12 pages
Pattern Matching - Elixir
No ratings yet
Pattern Matching - Elixir
5 pages
Adjoint Tutorial PDF
No ratings yet
Adjoint Tutorial PDF
6 pages
Chebyshev Polynomials
No ratings yet
Chebyshev Polynomials
19 pages
The Laplacian Operator From Cartesian To Cylindrical To
No ratings yet
The Laplacian Operator From Cartesian To Cylindrical To
7 pages
Numerical Method With Sage
No ratings yet
Numerical Method With Sage
19 pages
Finite Group Theory M. Aschbacher
No ratings yet
Finite Group Theory M. Aschbacher
316 pages
The Elements of Differentiable Programming
No ratings yet
The Elements of Differentiable Programming
300 pages
Roland Rudiger - Quantum Programming Languages:A Survey
No ratings yet
Roland Rudiger - Quantum Programming Languages:A Survey
46 pages
A Derivation of Einsteins Vacuum Field Equations
No ratings yet
A Derivation of Einsteins Vacuum Field Equations
24 pages
Plasma Turbulence in The Solar System PDF
No ratings yet
Plasma Turbulence in The Solar System PDF
107 pages
Banach Spaces
100% (2)
Banach Spaces
34 pages
David K. Ferry - 50 Years in The Semiconductor Underground (2015, Pan Stanford) PDF
No ratings yet
David K. Ferry - 50 Years in The Semiconductor Underground (2015, Pan Stanford) PDF
173 pages
Poincare Chapters 1&2
No ratings yet
Poincare Chapters 1&2
51 pages
A Tutorial On Algol 68
No ratings yet
A Tutorial On Algol 68
67 pages
Modern Cmake
No ratings yet
Modern Cmake
83 pages
Math EE IB
No ratings yet
Math EE IB
13 pages
Infinite Crossed Products
From Everand
Infinite Crossed Products
Donald S. Passman
No ratings yet
Introduction to the Theory of Error-Correcting Codes
From Everand
Introduction to the Theory of Error-Correcting Codes
Vera Pless
No ratings yet
Haskell Mini Reference: A Hitchhiker's Guide to the Modern Programming Languages, #10
From Everand
Haskell Mini Reference: A Hitchhiker's Guide to the Modern Programming Languages, #10
Harry Yoon
No ratings yet
A Survey of Minimal Surfaces
From Everand
A Survey of Minimal Surfaces
Robert Osserman
3.5/5 (1)
Substitutional Analysis
From Everand
Substitutional Analysis
Daniel Edwin Rutherford
No ratings yet
Group Theory in Solid State Physics and Photonics: Problem Solving with Mathematica
From Everand
Group Theory in Solid State Physics and Photonics: Problem Solving with Mathematica
Wolfram Hergert
No ratings yet
Genotyping Coronavirus SARS-CoV-2: Methods and Implications
No ratings yet
Genotyping Coronavirus SARS-CoV-2: Methods and Implications
12 pages
Rust Language Cheat Sheet
No ratings yet
Rust Language Cheat Sheet
9 pages
Anime Industry Report 2018
No ratings yet
Anime Industry Report 2018
8 pages
Sunny Weather Draws Crowds As Some Lockdowns Are Eased: Quotes
No ratings yet
Sunny Weather Draws Crowds As Some Lockdowns Are Eased: Quotes
28 pages
US Trade Deficit Rises Amid Record Drop in Exports: Quotes
No ratings yet
US Trade Deficit Rises Amid Record Drop in Exports: Quotes
28 pages
Rust Language Cheat Sheet
No ratings yet
Rust Language Cheat Sheet
9 pages
Dropbox Explained
No ratings yet
Dropbox Explained
9 pages
You Exec - Habit Tracker Part 2 Free
No ratings yet
You Exec - Habit Tracker Part 2 Free
232 pages
Oblique Shock Wave - Mechanical Engineering Department
No ratings yet
Oblique Shock Wave - Mechanical Engineering Department
9 pages
Science 9, Science 10, Mapeh 8, Tle 10
No ratings yet
Science 9, Science 10, Mapeh 8, Tle 10
2 pages
Document (Dental Casting Alloy
No ratings yet
Document (Dental Casting Alloy
11 pages
Method Acting Reconsidered Theory Practice Future 1st Ed 978 0 312 22309 0978 1 349 62271 9
No ratings yet
Method Acting Reconsidered Theory Practice Future 1st Ed 978 0 312 22309 0978 1 349 62271 9
302 pages
Animal Robot
No ratings yet
Animal Robot
11 pages
Lab 4
No ratings yet
Lab 4
2 pages
Focus Group Discussion Assessment Rubric
100% (1)
Focus Group Discussion Assessment Rubric
2 pages
LABS43 SBC Animal Feed
No ratings yet
LABS43 SBC Animal Feed
1 page
Din 1592
No ratings yet
Din 1592
1 page
Arun Resume 2022
No ratings yet
Arun Resume 2022
2 pages
Answer Key
No ratings yet
Answer Key
5 pages
The Rotordynamics Analysis of The Washing Machine Shaft Supported by Passive Magnetic
No ratings yet
The Rotordynamics Analysis of The Washing Machine Shaft Supported by Passive Magnetic
22 pages
Points East and West: Acupuncture and Comparative Philosophy of Science
No ratings yet
Points East and West: Acupuncture and Comparative Philosophy of Science
10 pages
Making A Modern Leader in Europe UK
No ratings yet
Making A Modern Leader in Europe UK
22 pages
Satip A 004 06
No ratings yet
Satip A 004 06
10 pages
Lecture Slides (Week 8) - Final Version
No ratings yet
Lecture Slides (Week 8) - Final Version
33 pages
CC1011 Midterm
No ratings yet
CC1011 Midterm
3 pages
InternshipReport 3
No ratings yet
InternshipReport 3
34 pages
Freewill and Determinism
No ratings yet
Freewill and Determinism
4 pages
Hdpe Boru Fiyat Listesi Hdpe Pipe Price List
No ratings yet
Hdpe Boru Fiyat Listesi Hdpe Pipe Price List
2 pages
Sand Reclamation and Conditioning
No ratings yet
Sand Reclamation and Conditioning
13 pages
EDA PP
No ratings yet
EDA PP
3 pages
SSL 38 Sresky
No ratings yet
SSL 38 Sresky
1 page
Cryptography in Blockchain
No ratings yet
Cryptography in Blockchain
24 pages
Comparative Stdy of Wind Analysis of Different Plan Configuration Building Using ETabs
No ratings yet
Comparative Stdy of Wind Analysis of Different Plan Configuration Building Using ETabs
18 pages
DIYEgg
100% (1)
DIYEgg
16 pages
Gerund Phrase 2
No ratings yet
Gerund Phrase 2
5 pages