Topology Explains Why Automobile Sunshades Fold Oddly

By Curtis Feist and Ramin Naimi, 2012

Annotated on Fermat's Library with 6 comments. Text and annotations as Markdown Original paper

About the authors: Curtis Feist completed his undergraduate and ...
This method is similar to how you would coil a bandsaw blade, which...
Braid theory in topology began to take shape in the early 20th cent...
The idea of the paper is that you can think of folding the car shad...
When a ribbon with a 360-degree twist is considered, the edges have...
A nice game to play with friends is to take a large ribbon and ask ...
arXiv:1205.4797v1 [math.GT] 22 May 2012
Topology Explains Why Automobile Sunshades
Fold Oddly
Curtis Feist and Ramin Naimi
Introduction
Most readers are probably familiar with Magic Shade “automatic folding”
sunshad es, even if they don’t recognize the brand name. These shades for
automobile windshields are roughly the shape of a circular disk when “open”,
and then fold up into a coil of several s maller circles when “closed”, thus
allowing the shades to be stored in a relatively small space.
Figure 1: The shade in its open (left) and closed (right) positions.
Let’s examine the s tructure of a Magic Shade a bit more closely. The key
part of the shade is a flat metal w ire that runs the length of the perimeter of
the sh ad e, forming a continuous loop. This wire is torsionally rigid. That is,
it doesn’t mind being bent along its length, but it does mind being twisted.
This observation w ill form the basis of our mathematical investigation, as
follows. The wire frame of a Magic Shade, when open, is in (roughly) the
shape of a single circle, with no twists in the wire. When closed, this wire
frame must be coiled into several smaller circles/loops, but without twisting
the wire. Specifically, we ask the following q uestions:
1
1. With the shade in its “normal” closed position, h ow m any lo op s does
the wire make, and why?
2. More generally, what are all possible closed positions for the shade, in
terms of how many loops the wire makes?
Note that the “how many” part of the first qu estion could be answered
by obser vation, but this would do nothing to address the “why” part of the
question.
First, a bit of terminology: Let’s call the flat wire fr ame a band. This
band has two edges, w hich we will call boundary circles. Let’s call the
circle midway between the boundary circles (see Figure 2) the core of the
band.
Figure 2: The band’s core (dotted curve) and boundary circles (solid curves)
with the shade in its open position.
Next, let’s discuss what we mean by saying that when the shade is in
its closed position, the band is “coiled” into several loops. Informally, this
would be similar to winding a length of ribbon around a s pool and then
joining the ends together (so that the piece of ribbon is itself a loop). In
fact, this is precisely what we have in m ind, except that for full generality
we also need to allow threading of the ribbon beneath loops that are already
on the spool.
Fortunately for us, the branch of topology known as braid theory (itself
a part of knot theory) exactly addresses this situ ation. That is (with the
shade in its folded position), we require the core of the band to be in braid
position, which we briefly summarize below.
2
Braids: A Brief Introduction
Informally, a braid is a set of one or more strands of strin g all starting
at the same height as each other, extending down without local minima or
maxima but with twisting allowed, and ending at the same height as each
other. Figure 3 shows an example. (For a more detailed introduction to
braid theory, see Adams [1].)
Figure 3: A four-strand braid.
A closed braid is a braid that is “closed up” by connecting the top
ends of the braid to its bottom ends in a pairwise fashion. Thus, the braid
is actually made up of one or more continuous loops of string, each of which
we call a component. For example, when the four-strand braid in Figure 3
is closed up, we end up with the two-component closed b raid in Figure 4(b).
The strands are drawn off to the side to avoid introd ucing any new crossings
into our diagram.
(a) (b)
Figure 4: (a) A one-component closed braid. (b) A two-component closed
braid.
In any (closed) braid one can easily arr an ge that whenever two strands
3
cross each other, no other pair of strands cross each other at that same
height. Thus, parallel horizontal lines can be drawn such that between any
two consecutive lines there is exactly one crossing, as in Figure 5. When
a closed braid is put in this configuration, we say it is in braid position.
Note that we have now made mathematically precise the idea of a ribbon
wrapped on a spool. Thus we have:
Statement 1. With the shade in its closed position, the core of the band is
a single-component closed braid in braid position.
Figure 5: Braid position: only one crossing in each “horizontal section”.
Now suppose our single-component closed braid was constructed from a
braid containing m strands and n crossings. It is not difficult to see that in
this case m and n must have opposite parity. Figure 4(a) shows a simple
example of this. We will not prove this fact here, but we will give a brief
justification for those who are familiar with the symmetric groups: T he n
crossings between the m strands translate into a sequence of n transpositions
in S
m
. Having a single-component closed braid translates into saying that
the product of these n transpositions gives an m-cycle in S
m
. And it is
known that in this case m and n must have opposite parity (Lanski [2,
pages 140-141, Theorems 3.13, 3.14] ). We now h ave:
Statement 2. With the shade in its closed position, if the band is coiled
into m loops and the core of the band (which is in braid position) has n
crossings, then m and n have opposite parity.
4
Now it’s time to consider the two boundary circles of the band. This
leads us naturally to another topic in topology, namely links and linking
number.
Linking Number: Another Brief Introduction
Informally, a link is a set of n continuous loops (again called components)
in 3-space. Figure 6 shows a few examples. (When n happens to be 1,
we also call our link a knot.) Note that a closed braid is simply a specific
presentation of a knot or link. (For a more detailed introduction to links,
see Adams [1].)
(a) (b)
(c)
(d)
Figure 6: Th ree two-component links, and one th ree-component link.
There is a very useful notion, called the linking number, that applies
to links with two components. Roughly speaking, it is the number of times
one of the components twists around the other. We will require a precise
definition, which we develop below.
Given a diagram of a 2-compon ent link , we first orient each component,
i.e., we put a “dir ection” on each component of the link (as if telling an
ant which way to travel). The direction we choose for each component is
arbitrary.
Now, we f ocus only on crossings between the two components (rather
than crossings between a component and itself). Each su ch crossing consists
of two oriented segments or “arrows”, as show n in Figure 7. We assign +1
or −1 to each crossing as follows: If the head of the top arrow is to the right
of the bottom arrow (when facing in the direction of the bottom arrow ), we
assign +1 to the crossing; otherwise we assign −1.
5
+1
-1
Figure 7: A right-handed (+1) and a left-handed (− 1) crossing.
To finish, we add up all the +1s an d −1s and then divide the sum by 2.
The result is called the linking number of the link for the given orientations.
Figure 8 shows an example of an oriented link with linking number 1.
+1
+1
+1
-1
Figure 8: Linking number = (1 + 1 + 1 − 1)/2 = 1.
What will be im portant f or us is the fact that the absolute value of the
linking number is a topological invariant of the link. That is, no matter
how the link is “moved aroun d,” what diagram of it we look at, and what
orientation we pick for each component, the absolute value of the linking
number will not change. The interested reader is referred to Adams [1] for
a proof of this fact. Now, with th e shade in its open position, the link ing
number of the boundary circles is clearly zero (see Figures 2 and 6(b)). Since
this linking number does not change when the shade is f olded up, we have:
Statement 3. With the shade in its closed position, the linking number of
the boundary circles is zero.
Analysis of the Magic Shade using Braid Position
and Linking Number
Let’s begin by recalling our firs t two Statements: With the shade in its
closed position, we assume that the core of the band is a single-component
6
closed braid in braid position; and if the band is coiled into m loops and the
core of the band has n crossings, then m and n have opposite parity.
Now draw in the two boundary circles, very close to the core. Since the
band has no twists in it, the only cr ossings we will see between the two
boundary circles are near the n s elf-crossings of the core, as in Figure 9.
The core has now s er ved its pu rpose; let’s take it out of th e picture. In
place of each of the n self-crossings of the core we now see four new crossings
involving the boundary circles: two self-crossings (between a boundary circle
and itself), and two non-self-crossings (between different boundary circles).
Since we’re interested in the linking number of the boundary circles, we
ignore the former type of crossing and concentrate on ly on the latter. So, we
have a total of 2n crossings, arran ged as n pairs, between the two boundary
circles.
Core
First boundary circle
Second boundary circle
S = Self-crossing
N = Non-self-crossing
N
N
S
S
Figure 9: Two self- and two non-self-crossings of the boundary circles near
one self-crossing of the core.
We compute the linking number of the two boundary circles as follows:
First, we give each boundary circle an arbitrary orientation and then assign
+1s and −1s to each of the 2n crossings. Next, recalling th at the core of
the band is in braid position, we note that each boundary circle is as well.
Thus, looking again at Figure 9, notice that for each boundary circle, its two
strands in that figure w ill be oriented either both downward or both upward.
It follows that in each of the n pairs of crossings, we h ave either two +1s or
two −1s, never one of each. Recall that the linking number is the absolute
value of half the sum of all these +1s and −1s; hence each of these n pairs of
crossings contributes exactly 1 or −1, never 0, to the linking number. Now,
7
+1 and − 1 are both odd, and if we add up n odd numbers, the result has
the same parity as n. But, by Statement 3, the link ing number is zero; so n
must be even. Hence, by Statement 2, m must be odd! That is:
Theorem. In its closed position, the wire frame of a Magic Shade sunshade
must be coiled into an odd number of loops.
At this point, if the reader has such a sunshade hand y, he or she should
verify that the wire frame in fact coils into three loops when the shade is
closed — thus answering Question 1 in the I ntroduction — and that ind eed
there are no twists in the wire. To answer Question 2 in the Introduction, the
reader may also verify that while it is possible to force the shade into two or
four coils, this will introduce twists in the wire, and the shade will certainly
not like being put in this position. But w hat about larger odd numbers? It
turns out that with a little trial and error, the shade can indeed be folded
into five (quite small) loops! The process for doing so actually involves a
common topology “trick”: With the shade closed into three loops, make two
of those loop s very small. Then close up th e big loop (as if it were an open
shade) into three loops, while the two smaller ones “go along for the ride”.
Following this idea, it is then clear that, at least theoretically, the shade
could be folded into any odd number of loops.
References
[1] C. Adams, The Knot Book: An Elementary Introduction to the Mathe-
matical Theory of Knots. W. H. Freeman and Company, New York, 1994.
[2] C. Lanski, Concepts in Abstract Algebra. Thomson Brooks/Cole, Bel-
mont, California, 2005.
8

Discussion

Tap the paper where you want to comment
About the authors: Curtis Feist completed his undergraduate and master's degrees at California Polytechnic State University in San Luis Obispo. He earned his Ph.D. in 1998 from the University of California, Davis, under Abigail Thompson's guidance. After serving as an assistant professor for three years at Marshall University in West Virginia, he has taken up the role of associate professor in mathematics at Southern Oregon University. ![](https://inside.sou.edu/assets/stem/math/images/Curtis_Feist_2010.jpg) Ramin Naimi received his Ph.D. from Caltech in 1992. Currently, he is a faculty member at Occidental College, where he dedicates himself to teaching and research while also enjoying mountain biking and playing squash. ![](https://www.oxy.edu/sites/default/files/styles/faculty_profile_image/public/faculty-images/naimi-ramin_400x400.jpg) This method is similar to how you would coil a bandsaw blade, which in this case has two easily identifiable edges. When coiled in the correct manner, the orientation of all the teeth is uniform throughout the coil. Here's a [video](https://www.youtube.com/watch?v=KRarn9st2bg) on how to do it. Braid theory in topology began to take shape in the early 20th century with the work of Emil Artin. Artin's foundational paper, published in 1925, introduced the concept of braids as a series of strands interwoven in a particular way and formalized the mathematical study of their properties. This work was influenced by earlier ideas from knot theory, which are similar to braids but closed into loops. Braids, however, are distinct in that their ends are not joined. Braid theory has since evolved, finding applications in various fields including cryptography, quantum computing, and the study of DNA. ![](https://miro.medium.com/v2/resize:fit:2000/1*3cP_BWx2lvf2XJ8ucjEQVw.png) The idea of the paper is that you can think of folding the car shade as winding a ribbon with the ends joined together around a cylinder without any twists. When a ribbon with a 360-degree twist is considered, the edges have a linking number of one, leading to an odd number of crossings and, consequently, an even number of loops. In this case, a twisted ribbon would typically form two loops. ![](https://jrh794.wordpress.com/wp-content/uploads/2010/12/twistedribbon2loops.jpg) With a 720-degree twist, the linking number becomes two, resulting in an odd number of loops. For a ribbon twisted three times, the linking number is three, leading back to an even number of loops. Here's a triply Twisted Ribbon with 4 loops ![](https://jrh794.wordpress.com/wp-content/uploads/2010/12/triplytwistedribbon4loops.jpg) [All the images were taken by Prof. Jim Hatton] A nice game to play with friends is to take a large ribbon and ask them around their arms with no twists with a given number of loops.