Graphically Speaking
Editor:
Miguel Encarnação
Beautiful Math—Aesthetic Patterns Based on
Logarithmic Spirals
Xinchang Wang
Jiangxi University of Finance and Economics
Peichang Ouyang
Jinggangshan University
S pirals’ self-similarity enables organisms
to maintain global harmony as they grow.
So, structures in many animals and plants
(such as shells, horns, and tree rings) have a spi-
ral appearance. On extreme scales, both galaxies
rize objects’ principal characteristics. For example,
a square’s symmetry group has two generators: a
90-degree rotation around the center and reflec-
tion on the diagonal. Every symmetry of a square
can be represented as the combined transforma-
and DNA molecules display a spiral structure. (For tions of these two generators.
more exploration of spirals in nature and spirals’ Finally, the fundamental region under a symme-
cultural and metaphysical meanings, see Spiral try group G is a connected set whose transformed
Symmetry.1) copies under the action of G cover the entire space
Spirals’ elegant, harmonious structure provides without overlapping, except at boundaries.
potential material for creating beautiful patterns. A logarithmic spiral is a curve that continually
To the best of our knowledge, the only paper on diminishes in curvature as it recedes from its cen-
generating spiral patterns is “Spiral Tilings with tral starting point. We can conveniently represent
Color Symmetry from Dynamics.”2 That paper es- such a spiral using polar coordinates:
tablished a relation between spiral symmetry and
wallpaper symmetry that you can use to produce , 1 ≠ a ∈ R+, q ∈ R,
spiral patterns indirectly.
Motivated by spirals’ aesthetic value, we devised where n is a positive integer and a is a positive real
a simple, fast method that exploits logarithmic number. As q increases, the spiral winds clockwise
spirals’ essential symmetries to create infinite ap- if a > 1 and counterclockwise if a < 1. Without
pealing patterns. losing generality, we assume a > 1 throughout this
article.
The Symmetry Group of Spirals The polar coordinate region W is this set:
To analyze spirals’ structure, we use the concept of
a symmetry group. We start with three definitions.
First, a symmetry of a pattern is a congruent .
or an isometric transformation (a common expla-
nation of symmetry is that a pattern is invariant
under this transformation). A pattern’s symmetry In Figure 1, the yellow region W enclosed by rays
group comprises all its symmetries. q1 = 0 and q2 = and spiral arcs r1 and r2 is the
Second, the elements g1, g2, ..., g n of a group fundamental region associated with the spiral’s
Q are a set of generators if you can express every symmetry group. The green region
element of Q as a finite product of their powers is the image of W under the transformation .
(including negative powers). Generators summa- The tiles , k = …, –1, 0, 1, …, form a spiral
Published by the IEEE Computer Society 0272-1716/13/$31.00 © 2013 IEEE IEEE Computer Graphics and Applications 21
Graphically Speaking
Given a pattern P with the symmetry group G,
a mapping M is invariant with respect to G if it
maps symmetrically placed points z and tz of P to
the same images. That is,
M(z) = M(tz), ∀ t ∈ G, z and tz ∈ P.
To produce a natural, appealing image, the
generated pattern should be seamless. This means
the dynamical system should be continuous on the
whole pattern. Dynamical systems satisfying this
requirement are the invariant mappings.
Given a spiral with W, let
with and . Then,
(r1, 0), (r2, 2p/n)(1)
Figure 1. A schematic of the symmetry of spiral patterns (ρ = a (2p/n)θ).
The yellow region W enclosed by rays q1 = 0 and q2 = and spiral arcs r 1 and
and r 2 is the fundamental region associated with the spiral’s symmetry
group. The green region Ω1 = a –2p/n (Ω) is the image of W under the (2)
transformation a –2p/n. If we successively apply generator a 2p/n, the
resulting images a k(2p/n) (Ω), k = …, –1, 0, 1, …, will form a spiral tiling. are, respectively, a pair of symmetrical points on
the rays q1 = 0 and q2 = 2p/n and arcs r1 and r2.
We define two functions:
tiling. From a mathematical viewpoint, is a
generator of G.
.
Colored Spiral Patterns from Dynamics
Dynamical systems, usually represented as equa- For the symmetrical points in Equations 1 and 2,
tions, describe how some quantities change over we can easily clarify that A and B have the same
time. Most dynamical systems are highly sensitive function values. Assuming that F with the initial
to the initial condition.3 We exploit that sensitiv- values A and B is a dynamical system, symmetri-
ity to produce colored spiral patterns. cally placed points of the spiral will have the same
Given a dynamical system F and a point (r0, iterative sequence. So, the resulting spiral pattern
q0) ∈ W, according to iterative sequences Fk(r0, q0) will be seamless.
(where k is a positive integer), we assign a color
to (r0, q0) according to a certain color scheme. A Gallery of Spiral Patterns
Owing to dynamical systems’ sensitivity, differ- Thus, we can select a variety of elementary functions
ent regions of W might present different colors. as invariant mappings. For example, to create Figure
Assigning color point by point, we can obtain a 3, the mapping M : R2 → R2 we used was
colored W.
For (r0, q0) ∉ W by symmetries, a proper trans-
formation (where m is a certain inte- .
ger) exists such that
(r0, q0) = y(r, q) ∈ W.
Because spirals are elegant, we can reconsti-
In this situation, we assign point (r, q) the same tute them to form other kinds of patterns. Fig-
color as (r0, q0). So, symmetrical points will dis- ure 2 shows a normal spiral pattern. Figure 3
play the same color, and the resulting patterns will illustrates a heart-like pattern we obtained by
have the desired spiral symmetries. arranging two copies of the same spiral pattern
To enhance visual appeal, we used the orbit trap face to face. Figure 4 demonstrates an interlock-
color scheme, which can help generate nice pat- ing pattern with half-turn symmetry. Figure 5
terns efficiently. Peichang Ouyang and his col- shows a pattern we constructed by interlocking
leagues employed it to color hyperbolic patterns.4 four spiral patterns.
22 November/December 2013
Figure 2. A normal spiral pattern based on ρ = Figure 3. A heart-like pattern constructed by
1.1(2p/18)θ. arranging two copies of the same spiral pattern
(ρ = 1.4(2p/15)θ) face to face.
Acknowledgments
We produced the patterns in this article in the
programming environment VC++ 6.0 with the aid
of the powerful graphics software package OpenGL.
We thank the Khronos Group and Microsoft for their
friendly technical support.
References
1. I. Hargittai and C.A. Pickover, Spiral Symmetry,
World Scientific, 1992.
2. K.W. Chung, H.S.Y. Chan, and B.N. Wang, “Spiral Tilings
with Color Symmetry from Dynamics,” Computers
& Graphics, vol. 23, no. 2, 1999, pp. 439–448.
3. M. Field and M. Golubitsky, Symmetry in Chaos,
Oxford Univ. Press, 1992. Figure 4. A half-turn pattern constructed by two spiral patterns
4. P.C. Ouyang et al., “The Visualization of Hyperbolic (ρ = 1.175(2p/21)θ ).
Patterns from Invariant Mapping Method,” Computer
& Graphics, vol. 36, no. 2, 2012, pp. 92–100. Figure 5.
A pattern
Xinchang Wang is a PhD student in the Information constructed by
Department of Jiangxi University of Finance and Econom- interlocking
ics and an associate professor at Jinggangshan Universi- four spiral
ty’s School of Mathematics and Physics. Contact him at patterns
wangxinchang11@[Link]. (ρ = 1.3(2p/20)θ).
Peichang Ouyang (the corresponding author) is a lecturer
at Jinggangshan University’s School of Mathematics and
Physics. Contact him at g_fcayang@[Link].
Contact department editor Miguel Encarnação at lme@
[Link].
Selected CS articles and columns are also available
for free at [Link]
IEEE Computer Graphics and Applications 23