Professional Documents
Culture Documents
(Series On Knots and Everything - Vol. 21) Slavik Jablan & Radmila Sazdanović-LINKNOT Knot Theory by Computer-WS
(Series On Knots and Everything - Vol. 21) Slavik Jablan & Radmila Sazdanović-LINKNOT Knot Theory by Computer-WS
The Series on Knots and Everything: is a book series polarized around the theory of
knots. Volume 1 in the series is Louis H Kauffman’s Knots and Physics.
One purpose of this series is to continue the exploration of many of the themes
indicated in Volume 1. These themes reach out beyond knot theory into physics,
mathematics, logic, linguistics, philosophy, biology and practical experience. All of
these outreaches have relations with knot theory when knot theory is regarded as a
pivot or meeting place for apparently separate ideas. Knots act as such a pivotal place.
We do not fully understand why this is so. The series represents stages in the
exploration of this nexus.
Details of the titles in this series to date give a picture of the enterprise.
Published:
Slavikjablan
Radmila Sazdanovic
The Mathematical Institute, Belgrade, Serbia
>World Scientific
NEW J E R S E Y • L O N D O N • S I N G A P O R E • BEIJING • S H A N G H A I • H O N G K O N G • T A I P E I • C H E N N A I
Published by
World Scientific Publishing Co. Pte. Ltd.
5 Toh Tuck Link, Singapore 596224
USA office: 27 Warren Street, Suite 401-402, Hackensack, NJ 07601
UK office: 57 Shelton Street, Covent Garden, London WC2H 9HE
For photocopying of material in this volume, please pay a copying fee through the Copyright
Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA. In this case permission to
photocopy is not required from the publisher.
ISBN-13 978-981-277-223-7
ISBN-10 981-277-223-5
Printed in Singapore.
Preface
Knot theory is a new and rich field of mathematics. Although “real” knots
are familiar to everyone and many ideas in knot theory can be formulated
in everyday language, it is an area abundant with open questions.
One of the main ideas of this book is to avoid obvious classification
of knots and links according to their number of components. For this
reason knots and links are referred to as KLs and treated together whenever
possible.
KLs are denoted by Conway symbols, a geometrical-combinatorial way
to describe and derive KLs. The same notation is used in the Mathematica
based computer program LinKnot that represents an integral part of this
book. LinKnot is not only a supplementary computer program, but the
best and most efficient tool for obtaining almost all of the results presented
in the book, that belong to the field of experimental mathematics.
Hands-on computations using Mathematica or the webMathematica
package LinKnot along with detailed illustrations facilitate better learn-
ing and understanding. The program LinKnot can be downloaded from
the web address http://www.mi.sanu.ac.yu/vismath/linknot/ and used as
a powerful educational and research tool for experimental mathematics– im-
plementation of Caudron’s ideas and the Conway notation enables working
with large families of knots and links. The electronic version of this book
and the program LinKnot that provides webMathematica on-line computa-
tions are available at the address http://math.ict.edu.yu/.
Each knot theory problem described in this book is accompanied with
the corresponding LinKnot function that enables the reader to actively
use the program LinKnot, not only for illustrating some problems, but for
computations and experimentation. LinKnot is software open to future de-
velopment: a reader can change it or add new functions. For the systematic
v
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
vi LinKnot
Contents
Preface v
1. Notation of Knots and Links 1
1.1 Basic graph theory . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Shadows of KLs . . . . . . . . . . . . . . . . . . . . . . . 10
1.2.1 Gauss and Dowker code . . . . . . . . . . . . . . . 16
1.3 KL diagrams . . . . . . . . . . . . . . . . . . . . . . . . . 25
1.4 Reidemeister moves . . . . . . . . . . . . . . . . . . . . . . 40
1.5 Conway notation . . . . . . . . . . . . . . . . . . . . . . . 50
1.6 Classification of KLs . . . . . . . . . . . . . . . . . . . . . 59
1.7 LinKnot functions and KL notation . . . . . . . . . . . . 66
1.8 Rational world and KL invariants . . . . . . . . . . . . . 69
1.8.1 Chirality of rational KLs . . . . . . . . . . . . . . 77
1.9 Unlinking number and unlinking gap . . . . . . . . . . . . 81
1.10 Prime and composite KLs . . . . . . . . . . . . . . . . . . 119
1.11 Non-invertible KLs . . . . . . . . . . . . . . . . . . . . . . 125
1.11.1 Tangle types . . . . . . . . . . . . . . . . . . . . . 131
1.11.2 Non-invertible pretzel knots . . . . . . . . . . . . 136
1.11.3 Non-invertible arborescent knots . . . . . . . . . . 140
1.11.4 Non-invertible polyhedral knots . . . . . . . . . . 142
1.12 Reduction of R-tangles . . . . . . . . . . . . . . . . . . . . 145
1.12.1 KLs with unlinking number one . . . . . . . . . . 148
1.13 Braids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
1.13.1 KLs and braids . . . . . . . . . . . . . . . . . . . 161
1.14 Braid family representatives . . . . . . . . . . . . . . . . . 165
1.14.1 Applications of minimum braids and braid family
representatives . . . . . . . . . . . . . . . . . . . . 179
vii
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
viii LinKnot
Contents ix
Bibliography 459
Index 475
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
Chapter 1
Definition 1.2. The valence (or degree) of a vertex is the number of edges
1
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
2 LinKnot
which are incident to it (in a graph with loops, we usually count a loop-edge
twice). The valence of a vertex v will be denoted d(v).
Since knots are 1-component links (c = 1), we will use the term “link”
or KL (knot or link) for both knots and links, unless we need to talk about
properties specific to knots only.
A graph is (3, 4)-valent if it contains vertices of valences only 3 or 4.
Four-valent graphs will be extremely important for study of KLs as they
represent KL shadows.
Among the graphs corresponding to five Platonic regular polyhedra, the
tetrahedron, cube and dodecahedron graphs are 3-valent, the octahedron
graph is 4-valent, and the icosahedron graph is 5-valent (Fig. 1.1).
K5 = {{1, 2}, {1, 3}, {1, 4}, {1, 5}, {2, 3}, {2, 4}, {2, 5}, {3, 4}, {3, 5}, {4, 5}},
K3,3 = {{1, 4}, {1, 5}, {1, 6}, {2, 4}, {2, 5}, {2, 6}, {3, 4}, {3, 5}, {3, 6}}.
Fig. 1.1 (a) Tetrahedron, (b) cube, (c) dodecahedron, (d) octahedron, (e) icosahedron
graph.
Definition 1.6. Two vertices are connected if there is a walk from one to
the other.
4 LinKnot
O = {{1, 2}, {1, 3}, {1, 5}, {1, 6}, {2, 3}, {2, 4}, {2, 6}, {3, 4}, {3, 5}, {4, 5}, {4, 6}, {5, 6}}
is
{{1, 2, 6, 5, 3}, {2, 3, 4, 6, 1}, {3, 1, 5, 4, 2}, {4, 3, 5, 6, 2}, {5, 3, 1, 6, 4}, {6, 2, 4, 5, 1}}.
After drawing the first vertex 1, we draw its incident edges in the right
cyclic order: {1, 2}, {1, 6}, {1, 5}, {1, 3}, then we continue with the vertex 2
and its adjacent edges in the same right cyclic order: {2, 3}, {2, 4}, {2, 6},
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
6 LinKnot
Fig. 1.2 Isomorphic graphs (a), (b), and their non-isomorphic duals (a′ ), (b′ ).
{2, 1}, having in mind that {2, 1} is already drawn as {1, 2}, etc., until
using all edges of the graph.
LinKnot function fPlanarEmbGraph gives the planar embedding of
a 3-connected planar graph given by a list of unordered pairs. The output
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
is a list that consists of the input graph, its planar embedding, and the
faces of the planar embedded graph. The basis of this program is the
external program planarity.exe written by J.M. Boyer (Boyer and Myrvold,
2005). The LinKnot function DrawPlanarEmbGraph draws a planar
embedding of a graph given by a list of unordered pairs, and the function
DrawPlanarEmbKL draws a planar embedding of a KL without multiple
edges. The basis of these functions is the program 3-Dimensional Convex
Drawings of 3-Connected Planar Graphs by M. Ochiai, N. Imafuji and N.
Morimura.
In every plane graph drawn in the plane ℜ2 we visually distinguish an
external face and internal faces placed inside it. In the octahedron plane
graph (Fig. 1.3) the external face is {1, 2, 3}, and other (internal) faces are
placed inside it. Sometimes, especially for the plane graphs obtained from
symmetric polyhedra, it is useful to imagine them on a sphere S 3 .
v − e + f = c + 1.
8 LinKnot
• • ←→ • • •
10 LinKnot
Knots have been used for various purposes from prehistoric times till today,
even serving as the basis for mathematical recording systems (e.g., for Inca
quipu). Examples of knots can be found in all ancient civilizations, in
Chinese art, Celtic art, ethnic Tamil and Tchokwe art, in Arabian, Greek
or Smyrnian laces... In contemporary science and art, KLs can be found in
DNA, physics, chemistry, sculpture, etc. Knots in design are the example of
modular structures (Jablan, 2002), since they can be composed from only
five basic pieces (modules) (Fig. 1.5).
Before giving a precise definition of KLs, we can start from an intuitive
description of knotting. Given a piece of string (like shoe laces or yarn)
we tangle it any way we want. A “mathematical” KL is different from
the “real” one since it is closed. A string ends should be glued together.
Moreover, mathematical KLs are made of string with no thickness– just a
closed curve in 3D-space with no self-intersections. A link is a set of several
disjoint tangled knots.
12 LinKnot
The set X together with a collection of open sets satisfying the rules
(1), (2), (3) is a topological space.
Theorem 1.5. Two knots K1 and K2 are ambient isotopic iff their com-
plements are homeomorphic (Gordon and Luecke, 1989).
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
However, this statement is not true for links; according to the famous
Whitehead example, two links that are not ambient isotopic can have home-
omorphic complements (Fig. 1.6a). Whitehead proved that there are infi-
nitely many links with the same complement as the Whitehead link (Fig.
1.6a, left) (Whitehead, 1937; Gordon, 2002).
In order to avoid the necessity of introducing differentiable or smooth
curves, as well to avoid some peculiar cases such as wild KLs (Fig. 1.6b),
we can think about KLs as piecewise linear.
All smooth KLs are tame, and the polygonal (piece-wise linear) knot
theory approach is equivalent to the smooth-curve approach (Crowel and
Fox, 1965; Burde and Zieschang, 1985).
For polygonal KLs a elementary planar isotopy is achieved either by
subdividing an edge AB by the vertex C, or by applying a contraction on
AC and CB. An ambient isotopy for a polygonal KL is a finite sequence
of elementary isotopies.
Fig. 1.6 (a) Different links with homeomorphic complements; (b) wild knot.
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
14 LinKnot
This means that every vertex of a shadow will be 4-valent, two or more
vertices of the polygonal link L can not be projected to the same point,
vertices of L′ can not belong to an interior point of any edge, three or more
points can not be projected to the same point, and projections of different
edges of L or their parts can not coincide in L′ . The forbidden situations
can be removed by a suitable choice of projection plane and/or by slightly
displacing the vertices of the polygonal link L. As with any other graph, a
shadow of a KL can be given by a list of unordered pairs, or by an adjacency
list, but the information from which you can draw regular shadow is given
by the code of planar embedding of its graph.
From the graph theory point of view, link shadow L′ is a 4-valent (or
4-regular) plane graph. The most important question is how much infor-
mation about link L can be obtained from its shadow L′ . As an example,
we will describe how to determine number of components of a link L from
its shadow L′ , according to the following rules:
Component Algorithm
Using all edges means that we have traced all components of L′ and
obtained one of the simplest KL invariants– the component number c.
From every 4-valent plane graph L′ there is a link L, such that L′ is its
shadow, and the number of circuits c in L′ obtained by the Component
Algorithm is the number of components of L.
The LinKnot function fComponentNo calculates the number of com-
ponents of any KL.
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
16 LinKnot
So far, all knot and link tables have been organized according to the
number of components of KLs. In order to make classification of KLs
and find common properties of knots and links, we will try to avoid this
approach as long as possible.
{{1, 2, 6, 5, 3}, {2, 3, 4, 6, 1}, {3, 1, 5, 4, 2}, {4, 3, 5, 6, 2}, {5, 3, 1, 6, 4}, {6, 2, 4, 5, 1}},
beginning from 1 and (1, 2), then from 1 and (1, 6), and finally from 2 and
(2, 6) we obtain the Gauss code
{{1, 2, 4, 5}, {1, 6, 4, 3}, {2, 6, 5, 3}},
so it is a shadow of well known 3-component link– Borromean rings 632 (Fig.
1.9).
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
The other way round, graph of a KL can be recovered from its Gauss
code. Unoriented edges of KL graph are given by unordered pairs of ad-
jacent numbers in each component of the Gauss code (where the first and
the last number in every component are also adjacent) (see Fig. 1.10).
The Gauss code of a KL is invariant with regard to a change of the
order of components, cyclic rotation and reversing components (if we are
not interested in the orientation of components).
According to the Component Algorithm, for every component we choose
a beginning point, and a first edge which induces the orientation of a whole
component. For a knot shadow with n crossings we have n choices for the
beginning point and 4 choices for direction, total of 4n possibilities. In the
case of links, this number grows with the number of components.
At the beginning of every classification the natural problem arises: to
enumerate all possible objects which can be obtained. Inspired by William
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
18 LinKnot
Thomson’s (Lord Kelvin’s) vortex theory (1867), his friend P.G. Tait
started with knot tabulation. He established cooperation with the Rev-
erend Thomas Penyngton Kirkman, who spend about 30 years considering
combinatorial problems in graph theory. Kirkman translated knot enumer-
ation to a problem of enumerating 4-regular planar graphs– shadows of
knots. In his first paper sent to Tait in May 1884 Kirkman enumerated
all knot projections with n ≤ 10 crossings. Tait used these tables to ex-
tract different alternating knots with n ≤ 10 crossings. Before publishing
their tables, Tait received an enumeration of knots up to 10 crossings from
C.N. Little, the material from his Ph.D. thesis On Knots, with a Census for
Order 10. After correcting one duplication in his own list, and a duplica-
tion and omission in Little’s, Tait sent the paper to press. In the meantime,
he received Kirkman’s list of 1581 knot projections with n ≤ 11 crossings
but decided that determining different knots is too demanding and retired
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
from knot tabulation in 1885 (Keller, 2004). T.P. Kirkman (1885a,b) and
P.G. Tait (1876-1885) completed the first knot tables of alternating knots
with n ≤ 10 crossings. Independently, after six years of work, C.N. Little
classified in 1889 non-alternating knots with n ≤ 10 crossings. Tait’s, Kirk-
man’s and Little’s tables of alternating knots are confirmed as complete,
and the only duplicate among 43 non-alternating knots with n = 10 cross-
ings is identified in 1974 (Perko pair). At the time when they finished their
work, after the wide recognition of D. Mendeleev’s periodic table of ele-
ments by scientific community, almost none was interested in knot tables.
After 20 years, mathematicians recovered the subject.
Before trying to repeat and extend Tait’s, Kirkman’s and Little’s results
(now using computers), it will be useful to try to reduce the number of
possibilities by minimizing the number of crossings, and to make the codes
of KLs as concise as possible.
Similar to the definition of a proper (or reduced) graph, a KL
shadow is called proper, or reduced if it has no loops. The Gauss code
{{1, 2, 3, 4, 2, 3, 4, 1}} represents a possible knot shadow (you can recognize
the shadow of a trefoil with a loop) (Fig. 1.11). In order to delete loops
from KL codes, first we need to recognize them. In a Gauss code, the
appearance of the same numbers in adjacent places indicates a loop (where
the first and the last number in any component are treated as adjacent as
well).
20 LinKnot
enumerate the second, beginning from its point that is visited at the most
once. If the parity is disturbed for some component, we continue from the
next number (that is the same as a cyclic rotation of a component used
before). Applying the same rule until two numbers, one odd and the other
even, are assigned to every crossing, we obtain a Dowker code of a KL (Fig.
1.13a).
Dowker code of a KL is sufficient for drawing its corresponding shadow
(Fig. 1.13b).
Fig. 1.13 (a) Obtaining the Dowker code of figure-eight knot from its shadow; (b)
drawing the shadow from the Dowker code of the same knot.
22 LinKnot
applied to links. For example, the Dowker code {{1, 1}, {4, 2}} denotes a
shadow of a Hopf link 221 (2 in Conway notation) without loops.
Now we have all necessary definitions and notions to pursue Kirkman’s
approach to systematic classification and enumeration of all alternating
knot diagrams with n crossings. In solving this problem, Dowker and Gauss
codes will be used.
In order to classify all KLs with n crossings first we create all distinct
permutations of even numbers 2,4, . . ., 2n and all their partitions into 1, 2,
. . ., n parts and obtain all possible Dowker codes with n crossings. Then
we delete from them non-proper codes. The result obtained are all possible
potential Dowker codes of KL shadows with n crossings. We emphasize
“potential” because not all of them are necessarily realizable. For example,
it is impossible to draw the potential Dowker code {{5}, {8, 10, 2, 4, 6}}
(Fig. 1.14a), since it is not planar graph. In fact, it is K5 . Moreover, a
Dowker code can be non-realizable even if its corresponding graph is planar.
For example, to the non-realizable Dowker code {{6}, {4, 6, 8, 10, 12, 2}}
corresponds the (non-realizable) Gauss code {1, 2, 3, 1, 4, 3, 5, 4, 6, 5, 2, 6},
and the planar graph
{{1, 2}, {1, 3}, {1, 4}, {1, 6}, {2, 3}, {2, 5}, {2, 6}, {3, 4}, {3, 5}, {4, 5}, {4, 6}, {5, 6}}
which is realizable as the link {{2, 2, 2}, {6, 8, 10, 12, 2, 4}}.
A bit strange, but a natural question is: where can a non-realizable
Dowker code be realized? The answer is: on some surfaces other
than plane ℜ2 (or sphere S 2 ). The graph K5 corresponding to the
code {{5}, {8, 10, 2, 4, 6}} can be embedded on a torus, and represents a
shadow of a 3-component link (Fig. 1.14b). One of its Gauss codes is
{{1, 5, 3}, {2, 5, 4}, {1, 2, 3, 4}}, but its Dowker code does not exist (because
Jordan Curve Theorem holds only in ℜ2 or S 2 ). In fact, this Gauss code
represents Borromean rings.
The virtual knot theory introduced by L. Kauffman (Kauffman, 1997,
1999, 2000, 2001; Green, 2004; Manturov, 2002, 2003, 2004; Zin-Justin
and Zuber, 2004; Zinn-Justin, 2006) is a “non-realizable” part of the knot
theory and gives the alternative answer to the question about realizability
of Dowker codes.
By projecting four-valent graphs onto ℜ2 or S 2 , virtual crossings are
intersection points in the projection which are not vertices of the original
graph. For example, graph on a torus with one vertex corresponds to the
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
Fig. 1.14 (a) An attempt to draw a knot shadow defined by the potential Dowker code
{{5}, {8, 10, 2, 4, 6}}; (b) the result obtained by applying the Component Algorithm on
the embedding of non-planar graph K5 on a torus (that we obtain by identifying opposite
edges of the square); (c) virtual links obtained from Hopf link and Borromean rings.
24 LinKnot
Fig. 1.15 (a) The graph of figure-eight knot; (b) the graph of Borromean rings.
1.3 KL diagrams
26 LinKnot
28 LinKnot
Fig. 1.20 The mirror image of the trefoil knot diagram from Fig. 1.19.
D = {{{1, 6}, {3, 8}}, {{5, 12}, {7, 10}}, {{9, 2}, {11, 4}}},
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
Fig. 1.21 Two non-isomorphic projections of the same knot 75 (Figs. 1.21, 1.22).
30 LinKnot
Fig. 1.22 Another 7-crossing projection of the knot 75 from Fig. 1.21.
(Fig. 1.23). These tables list knots classified according to their crossing
number. Every knot is presented by one minimal projection and denoted by
ordering number (in the Alexander-Briggs notation, 1926-27). It’s unclear
how Reidemeister selected minimal projections of those knots which have
more then one such. On the list of n-crossing knots, for each n, alternating
knots precede non-alternating knots. Most of the knot theory books closely
follow Reidemeister tables: knot projections are just redrawn, sometimes
turned upside down (e.g., 76 ), and never changed into a different minimal
projection of the same knot. The most influential book containing extensive
tables of KLs, and the first to use Conway notation from his seminal paper
(1970), is D. Rolfsen’s Knots and Links (1976). The only duplicate in his
tables was found by K. Perko (the famous Perko pair, Fig. 1.26), who
also corrected Conway’s eleven crossing knot tables, where four knots were
omitted (Perko, 1974, 1982).
and
respectively. In the same way, for the first non-alternating link, denoted
633 , Gauss codes can be
and
32 LinKnot
be denoted by a Dowker code with only positive numbers, and for non-
alternating KLs some entries will be negative. For example, the alternating
projection of the figure-eight knot and its mirror image will have the same
code {{4}, {6, 8, 2, 4}}; the non-alternating projection of the knot 819 (Fig.
1.17) and its mirror image has the code {{8}, {12, 14, 10, −16, 4, 6, 2, −8}}.
Likewise, the code {{2, 2, 2}, {8, −10, 2, 12, 6, −4}} is assigned to the non-
alternating projection of the link 633 (Fig. 1.18) and its mirror image. The
codes obtained will be called DT-codes (where DT comes from Dowker-
Thistlethwaite), or Dowker codes in Knotscape notation (according to the
computer program Knotscape where that notation is used).
It is important to underline that DT-codes essentially differ from a
Dowker code with signs that we will introduce now.
Fig. 1.24 (a) A vertex with the sign −1; (b) a vertex with the sign +1.
Fig. 1.25 Two different composite knots with the same Dowker codes.
34 LinKnot
The proof follows directly from the proof of the Kauffman-Murasugi The-
orem (Theorem 1.6), or the Tait’s Flyping Theorem (Theorem 1.11).
The following theorem is the consequence of the Theorem 1.7.
36 LinKnot
the Dowker code with signs of a KL given by its Dowker code in Knotscape
form (or DT-code). For an alternating KL, an input is the Dowker code
without signs, and for a non-alternating KL, an input is the Dowker code
containing only signs of crossings with signs changed with regard to the
corresponding alternating KL. The output is a Dowker code with signs.
The function fKnotscapeDow calculates from a Conway symbol of a KL
its Dowker code in the Knotscape format: Dowker code without signs for
an alternating KL, or Dowker code with signs of changed crossings for a
non-alternating KL.
The most complete, and at the same time, the most concise
code is P-data, which has the same form as a Dowker code. P -
data gives a numerical code, signs of crossing points, and for a non-
alternating KL the information about crossings with relation “over-under”
changed with regard to the corresponding alternating KL. The pro-
gram Knot 2000 (K2K) uses P -data to internally represent a KL. Let
us suppose that we have already computed for some KL its Dowker
code with signs Dow, and its DT-code. For example, for the non-
alternating knot 819 (Fig. 1.17), Dow={{8}, {12, 14, 10, 16, 4, 6, 2, 8}}, and
DT={{8}, {12, 14, 10, −16, 4, 6, 2, −8}}. From Dow we can obtain a list of
ordered pairs, where each crossing is labelled by two numbers: odd and
even. In our example, that list is:
D = {{1, 12}, {3, 14}, {5, 10}, {7, 16}, {9, 4}, {11, 6}, {13, 2}, {15, 8}}.
If in DT the k th number (k = 1, . . ., n) is negative, we need to reverse the
order of the numbers of the k th ordered pair in D, leaving signs in D at
their places. Because the positions of negative numbers in DT are {4, 8},
after reversing the 4th and 8th ordered pair in D, we obtain the list
{{1, 12}, {3, 14}, {5, 10}, {16, 7}, {9, 4}, {11, 6}, {13, 2}, {8, 15}}.
After sorting this list according to the first members of ordered pairs, we
obtain the list
D1 = {{1, 12}, {3, 14}, {5, 10}, {8, 15}, {9, 4}, {11, 6}, {13, 2}, {16, 7}}.
The first part of the P-data is the same as the first part of Dow, and
the second part, called P-word, is the list of the second members of the
ordered pairs from D1 , so P-data={{8}, {12, 14, 10, 15, 4, 6, 2, 7}}. In fact,
P-data is very similar to a Dowker code with signs. The only difference
is that P -data contains odd instead of the corresponding even numbers in
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
all crossings where the relation “over-under” is changed with regard to the
corresponding alternating KL.
Dowker codes contain all relevant information about a KL diagram, so
can be used as an input for computer programs. Morwen Thistlethwaite
used the Dowker notation to list all prime knots with up to 16 crossings
(Hoste, Thistlethwaite, and Weeks, 1998), and Jim Hoste derived the list
of all links with at the most nine crossings, representing each KL by its
minimal Dowker code (Doll and Hoste, 1991; Cerf, 1998).
Any regular diagram of a KL has a finite number of crossings, and this
number is called the crossing number of the diagram. Given all regular
diagrams of a KL, the crossing number of the regular diagram with the
fewest number of crossings is the crossing number of a KL (Definition
1.25). As we already mentioned, according to Kauffman-Murasugi Theorem
(Theorem 1.6) crossing number of an alternating KL is the crossing number
of its reduced alternating diagram.
Since every KL diagram can be presented with many different se-
quences, there are more sequences then KLs. If we are working with min-
imal diagrams of a KL (where the number of crossings coincides with the
crossing number of the KL) this ratio is finitely many to one. Otherwise,
we have infinitely many sequences for one KL.
The first attempt to minimize the amount of data is a minimization
of Dowker codes. We already mentioned that a Dowker code of a KL
projection is dependent on the choice of the beginning point of each com-
ponent and on its orientation (i.e., on the choice of the first oriented edge).
Among all Dowker codes which correspond to a specific KL projection we
can choose the minimal one. In the case of knots, this means choosing
the minimal permutation among all possible Dowker codes taken without
signs. In the case of links, in order to obtain the minimal Dowker code we
use two criteria in the following order: the length of components (where
shorter components have the priority), and the minimal code (i.e., minimal
permutation criterion).
The simplest, but certainly the slowest minimization algorithm cre-
ates all possible Dowker codes for a given knot projection, sorts them and
chooses the first: the minimal Dowker code of the given projection. In a
similar way, for a link projection one can make all possible choices for begin-
ning points of components and all their orientations, calculate all Dowker
codes and take the minimal one. A more smarter algorithm first sorts com-
ponents according their lengths, then finds an optimal beginning point for
each of them, permute components equivalent with regard to the two cri-
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
38 LinKnot
teria mentioned above, calculates Dowker codes and chooses the minimal
one. In this way, we can compute the minimal Dowker code for every KL
projection.
Sometimes it is useful to work with weighted graphs of KLs given by a
list of unordered pairs and list of vertex signs instead of Gauss or Dowker
codes. The LinKnot function fGraphInc calculates from a Conway sym-
bol, Dowker code, or P -data of any KL its corresponding graph. An output
is the graph given by edges (as a list of unordered pairs) and by the list of
vertex signs.
The LinKnot function fPlanarEmb calculates the planar embedding
of a prime KL given by a Conway symbol, P -data or Dowker code. An
output is the list that consists of the graph of the input KL, its planar
embedding given by vertex cycles, and the faces of the planar embedded
graph. The basis of this program is the external program planarity.exe
written by J.M. Boyer (Boyer and Myrvold, 2005).
As we already mentioned, every KL shadow is a 4-valent graph. If
we have any polyhedral graph G, we can obtain its corresponding mid-
edge graph M (G) defined by mid-edge points of G by connecting mid-edge
points belonging to adjacent edges of G. Clearly, the result M (G) is always
a 4-valent graph. For example, for the tetrahedron graph
{{1, 2}, {1, 3}, {1, 4}, {2, 3}, {2, 4}, {3, 4}}
{{1, 2}, {1, 3}, {1, 4}, {1, 5}, {2, 3}, {2, 4}, {2, 6}, {3, 5}, {3, 6}, {4, 5}, {4, 6}, {5, 6}}
Fig. 1.27 (a) Mid-edge graph obtained from tetrahedron graph; (b) figure-eight knot
reconstructed from its graph.
Two other graph functions work with graphs that are not necessarily
4-valent. The function fKLinGraph gives all non-isomorphic KL projec-
tions contained in a given graph G. For example, the graph (Fig. 1.28a)
{{1, 2}, {1, 2}, {1, 4}, {1, 4}, {1, 5}, {2, 3}, {2, 3}, {2, 5}, {3, 4}, {3, 4}, {3, 5}, {4, 5}}
{{1, 2}, {1, 2}, {1, 4}, {1, 4}, {2, 3}, {2, 3}, {3, 4}, {3, 4}}
{{1, 2}, {1, 2}, {1, 4}, {1, 5}, {2, 3}, {2, 5}, {3, 4}, {3, 4}, {3, 5}, {4, 5}}
{{1, 2}, {1, 3}, {1, 6}, {2, 4}, {2, 6}, {3, 4}, {3, 5}, {4, 5}, {5, 6}}
40 LinKnot
{{1, 2}, {1, 2}, {1, 3}, {1, 6}, {2, 4}, {2, 6}, {3, 4}, {3, 4}, {3, 5}, {4, 5}, {5, 6}, {5, 6}}
and
{{1, 2}, {1, 3}, {1, 3}, {1, 6}, {2, 4}, {2, 4}, {2, 6}, {3, 4}, {3, 5}, {4, 5}, {5, 6}, {5, 6}}.
Fig. 1.29 Graphs derived from 3-valent graph (a) by replacing some single edges by
double edges (b).
The next step in derivation of all KLs with a given number of crossings is
finding all different (non-isomorphic) minimal projections of a given KL,
i.e., all its different projections with the number of vertices equal to the
crossing number. In the case of alternating KLs it is sufficient to find all
proper non-isomorphic alternating projections of a given KL with a fixed
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
42 LinKnot
The authors emphasize that this function fails for some classes of links, but
most of its reductions are the optimal ones1 .
Before a more detailed discussion of this problem, we will describe the
elementary steps of a reduction process: Reidemeister moves2 . So far we
had only static images: KL shadows and projections. Now we can make
a movie with ambient isotopy playing the main role. All KLs will be
represented as polygonal KLs, and all the moves that consist of a finite
series of elementary isotopies will be expressed as finite compositions of
Reidemeister moves. The move Ω0 was already introduced as a planar
isotopy. Recall that for a polygonal link a planar isotopy Ω0 is achieved
either by subdividing an edge AB by the vertex C, or contracting AC and
CB. The move Ω0 can be introduced as the elementary planar isotopy
(Fig. 1.30). An ambient isotopy for a polygonal KL is a finite sequence of
elementary isotopies. Reidemeister moves, Ω1 , Ω2 , and Ω3 are illustrated in
Fig. 1.30. We represent Reidemeister moves as polygonal moves, and the
piecewise-linear and the smooth knot theory give the same classification of
KLs (Crowel and Fox, 1965; Burde and Zieschang, 1985). The equivalent
of ambient isotopy of KLs, for knot and link diagrams are Reidemeister
moves. Hence, Reidemeister moves are planar isotopies.
For the elegant proof of this theorem see Kauffman (2004) or Manturov
(2004, Theorem 2.1). The last book (appendix A) contains the proof of the
independence of Reidemeister moves.
1 This holds even for much better heuristic program for reduction of knots knotfind.c
written by M. Thistlethwaite and used as a part of the program Knotscape. Its first
known unsuccessful reduction is for a 40-crossing knot.
2 J.C. Maxwell determined all regions bounded with fewer then four arcs corresponding
to Reidemeister moves. The proof that they suffice to pass between equivalent diagrams
was published by both Reidemeister (1926) and Alexander and Briggs (1926-27) (Hoste,
2006).
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
Fig. 1.31 (a) A tangle, (b) flype; (c) mutation; (d) vertical mutation; (e) horizontal
mutation.
44 LinKnot
(1) for a given number of crossings n find all non-alternating knots having
at least two minimal diagrams with a different writhe;
(2) how many different values can writhe of an non-alternating knot take
on its minimal diagrams;
(3) is it possible to find two minimal diagrams of the same KL whose
writhes differ by an arbitrary number;
(4) can we transform one minimal diagram of a prime knot to another
minimal diagram with the same writhe using only Reidemeister moves
Ω2 and Ω3 which does not change a writhe?
46 LinKnot
Fig. 1.32 Two diagrams of the knot 2 2#8∗ with the same writhe and different winding
number.
48 LinKnot
Fig. 1.35 (a), (b) Nasty unknot; (c) Goeritz’s unknot; (d) Monster unknot.
50 LinKnot
Fig. 1.36 Hard (a,b) unknot and (c) unlink diagrams with n = 9 crossings.
The remaining two versions of the third move (Fig. 1.38b) are forbidden.
Actually, the forbidden move is a very strong one: each virtual knot can be
transformed to another one using all generalized Reidemeister moves and
the forbidden move (Nelson, 2001).
Fig. 1.37 Hard (a) unknot and (b) unlink diagrams with n = 10 crossings.
52 LinKnot
Fig. 1.38 (a) Generalized Reidemeister moves Ω′1 , Ω′2 , Ω′3 , Ω′′
3 ; (b) forbidden move.
tential of the Conway notation in the search for the universal classification
principle– “periodic table” of KLs. It is very surprising that the Conway
notation, the only geometrical-topological notation that gives complete,
interpretable and understandable information on KLs is still not widely
accepted.
Maybe the main reason why most of knot theory books have only the
classical notation is the non-uniqueness of the Conway notation. Hence, we
need to make a choice of the “standard” Conway symbol of a KL, accord-
ing to the notation introduced in the original Conway’s paper (Conway,
1970) and in the papers and books following it (Caudron, 1982; Rolfsen,
1976; Adams, 1994). Notation becomes more complicated for polyhedral
KLs. We need to specify the symbol of the particular basic polyhedron
and the particular order and orientation of its vertices (Figs. 2.43-2.47).
For example, the same link .2 can be denoted as : 2, : .2, :: 2, :: .2, or even
as 6∗ 2, 6∗ .2, 6∗ : .2, 6∗ :: 2, and 6∗ :: .2, where as the standard symbol we
choose the first of them.
The LinKnot function fClassicToCon gives Conway symbol of a KL
given in the classical notation.
Elementary tangles are shown in Fig. 1.39 and denoted by 0, 1 and −1,
where for alternating KLs 0 and 1 are sufficient. Any tangle can be obtained
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
54 LinKnot
Closed tangle can be obtained from a tangle in two ways (without in-
troducing additional crossings): joining in pairs NE and NW, and SE and
SW ends of a tangle we obtain a numerator closure; joining in pairs NE and
SE, and NW and SW ends we obtain a denominator closure (Fig. 1.43a,b).
Fig. 1.43 (a) Numerator closure; (b) denominator closure; (c) basic polyhedron 1∗ .
plats. O. Simony (1882, 1884) was the first who considered rational KLs
from the mathematical point of view, while their complete classification
was given by H. Schubert (1956). J. Conway noticed the beautiful relation
between rational tangles and continued fractions (Conway, 1970; Kauffman
and Lambropoulou, 2002) (Theorem 1.18).
The basic polyhedron 1∗ is illustrated in Fig. 1.43c, and the other basic
polyhedra with n ≤ 12 crossings in Figs. 2.43-2.47.
Three operations with tangles: sum, product, and ramification (page 52)
are sufficient for the notation of algebraic KLs. Polyhedral KLs require
special notation.
In addition to the operations used for algebraic KLs we need to know
a symbol of a basic polyhedron P ∗ = nO∗ , where n denotes the number
of vertices, and O the ordering number of a particular basic polyhedron
among those with the same number of vertices. For example, 123∗ is the
third basic polyhedron with n = 12 vertices. A knot or link obtained from a
basic polyhedron P ∗ =nO∗ by substituting tangles t1 , . . ., tk in appropriate
places is denoted by P ∗ t1 . . . tk , where the number of dots between two
successive tangles shows the number of omitted substituents of value 1.
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
56 LinKnot
Fig. 1.44 A link shadow collapsing into the basic polyhedron 6∗ – an octahedron.
Fig. 1.45 Basic polyhedron 6∗ and the knots 6∗ 2.1.2.1.2 0.1 and 6∗ 2 1.2.3 2 : −2, 2 0.
Conway symbols are used for the first time as an alternative notation
in the book Knots and Links by D. Rolfsen (1976), and then by some
other authors (e.g., C.C. Adams, 1994). Notice that in Rolfsen’s tables
some drawings of KLs do not correspond to their Conway symbols. For
example, in the case of the knot 915 , it is clear that from its Conway symbol
2 3 2 2 we obtain the projection with 5, and not with 4 bigons, pretzel knot
819 with the Conway symbol 3, 3, −2 is drawn as 6∗ 2 0. − 2 0. − 1. − 1, etc.
It is interesting that in Conway symbols of all non-alternating polyhe-
dral KLs with n ≤ 10 crossings the symbol . − 1, i.e., a single vertex with
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
a changed sign never appears, except in the case of the 4-component link
103∗ − 1. − 1. − 1. − 1 :: . − 1 (or 10−∗∗∗ according to Conway’s paper).
The first polyhedral non-alternating knots that can not be expressed by
Conway symbols without . − 1-s are 12-crossing knots
8∗ 2 0.−2.−1.−2 0.2 0, 8∗ 2 0.−2 0.−1.−2 0.2 0, 1212∗ −1.−1.−1.−1.−1.−1
(12n801 , 12n835 , and 12n837 in the Knotscape notation, respectively).
In the program Knot 2000 (K2K), the function GetPdatabyTracking
is used for entering KLs by drawing them in the mouse-tracking window.
The output is P -data of the KL.
Instead of the graphical input, LinKnot function fCreatePData uses
a Conway symbol of KL (given as a Mathematica string) and computes
P -data.
For example, the figure-eight knot 41 is denoted by ”2 2”, knot 95 by
”5 1 3”, link 521 is denoted by ”2 1 2”, link 9221 by ”3 1, 3, 2” (for all of them
a space between tangles denotes a product of tangles), etc. A sequence of k
pluses at the end of the Conway symbol is denoted by +k, and a sequence of
k minuses by + − k (e.g., knot 1076 given in Conway notation as 3, 3, 2 + +
is denoted by ”3, 3, 2 + 2”, and the link 9317 given in Conway notation as
3, 2, 2, 2 − − by ”3, 2, 2, 2 + −2”). The space denoting a product of tangles
is used in the same way in all other symbols. For example, the knot 10133
is denoted by ”2 3, 2 1, 2 + −1”, and the knot 10154 by ”(2 1, 2) − (2 1, 2)”
(with spaces).
The program LinKnot contains the database of basic polyhedra with
at most n = 20 crossings, where every basic polyhedron is represented by
its corresponding alternating KL diagram. For the basic polyhedra with
n < 10 crossings, the standard notation is used (.1, 6∗ , 8∗ , 9∗ , where sym-
bols of polyhedral KLs beginning with a dot correspond to Conway’s basic
polyhedron 6∗∗ or .1). For example, the knot 1095 is denoted by ”.2 1 0.2.2”,
and 10101 by ”2 1..2..2”. For n ≥ 10 in each symbol the first two digits rep-
resent the number of crossings, and the next the ordering number of the
polyhedron (e.g., 101∗, 102∗, 103∗ for n = 10 denoting 10∗ , 10∗∗ , 10∗∗∗ ,
respectively, and 111∗ , 112∗ , 113∗ for n = 11 denoting 11∗ , 11∗∗ , 11∗∗∗ ,
respectively, etc.). For n = 12 basic polyhedra are ordered according to
their list made by A. Caudron (1982), so polyhedra originally denoted with
12A-12L are 121∗ -1212∗. For n > 12 the database of basic polyhedra is
derived from the list of simple 4-regular, 4-edge-connected, but not neces-
sarily 3-vertex connected plane graphs generated by Brendan McKay using
the program plantri written by Gunnar Brinkmann and Brendan McKay
(http://cs.anu.edu.au/∼bdm/plantri/).
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
58 LinKnot
Fig. 1.46 Alternating links (a) 11∗∗∗ 2 ∼ 125∗ ; (b) 136∗ ∼ 1318∗ .
Definition 1.46.
60 LinKnot
informal term used to describe a list of knots where each successive knot
is obtained from the previous one by a simple process”. The source link 2
(i.e. the Hopf link 221 ) and its family are shown in Fig. 1.47.
62 LinKnot
This means that all KLs generated from a source link S by substituting
single bigons by chains of bigons make a family generated from S.
This means that a subfamily will be obtained if all ka from the Definition
1.48 are even.
Theorem 1.14. KLs belonging to the same subfamily have the same num-
ber of components.
The proof of this theorem is trivial, because the addition of any chain
of bigons of an even length to a bigon or chain of bigons does not change
the number of components.
If there is no chance for a confusion, in order to simplify, we will use
the common term “family” for both families and subfamilies.
According to experimental results, various KL invariants will be related
to subfamilies, and will represent some combination of numbers denoting
chains of bigons (integer tangles n or −n) from a general Conway symbol
of a subfamily.
The proof of this theorem is trivial as well. In the case of writhe, every
addition of two bigons to a chain of bigons changes writhe by +2 in the case
of positive, or by −2 in the case of negative chain, so knowing the writhe of
the generating alternating knot given by its (minimal) Conway symbol, it
is possible to calculate writhe of any knot belonging to a subfamily derived
from it. In the case of linking number, we need to take in account only
chains of bigons belonging to different components.
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
64 LinKnot
Fig. 1.48 A knot shadow with the symmetry group G = [2, 8] and its antisymmetry
subgroup G/G′ = [2, 8]/[2, 8]+ obtained by introducing the relation “over-under”.
Fig. 1.49 Graphs corresponding to (a) rational; (b) stellar; (c) arborescent KLs; (d)
four graphs of the same non-alternating knot 2 2, 3, −2; (e) graph transformation rules
(Caudron, 1982).
66 LinKnot
The program Knot 2000 (K2K) was created by Mitsuyuki Ochiai and Noriko
Imafuji from the Graduate School of human culture, Nara Women’s Univer-
sity, Nara, Japan. In 2003 the Mathematica-based program LinKnot written
by Slavik Jablan and Radmila Sazdanović from the Mathematical Institute,
Belgrade, Serbia, was combined with it. LinKnot is compatible with Knot
2000 and provides tools for working with KLs given in Conway notation
with no restriction on the number of crossings. The webMathemathica pro-
gram LinKnot providing on-line computations and the electronic version of
this book is available at the address http://www.math.ict.edu/.
An input for the program K2K is a KL diagram drawn by mouse on
the mouse tracking window (the function GetPdatabyTracking). The
LinKnot function fCreatePData gives the possibility to use a Conway
symbol of a KL as an input. For example, from the Conway symbol of
the non-alternating link K=”111 ∗ 2.2.2. − 2 0.2 0. − 2 0” it calculates the
corresponding P -data
{{14, 3}, {−16, −20, −12, 23, 29, −6, −30, 32, −2, −24, −4, 7, −18, 14, −28, 26, 9}}.
P -data is a list having two entries. The first is a list of the numbers of cross-
ings in each component, and the second is a list of numbers derived from
the KL, called P-word. P -data are the basic input for all K2K functions.
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
K = ”101∗2 1.2 1 0.2 1.2 1 0.2 1.2 1 0.2 1.2 1 0.2 1.2 1 0”
{{14}, {15, −21, −23, −27, −25, −19, −17, 1, −13, −3, −5, −11, −9, −7}}.
68 LinKnot
Fig. 1.51 Drawing of the link 111∗ 2.2.2. − 2 0.2 0. − 2 0 by using the functions fCre
atePData and ShowKnotfromPdata.
Fig. 1.52 Drawing of the link 101∗ 2 1.2 1 0.2 1.2 1 0.2 1.2 1 0.2 1.2 1 0.2 1.2 1 0.
{{14, 3}, {16, 20, 12, −22, −34, 6, 30, −32, 2, 24, 4, −8, 18, −14, −10, 28, −26}}.
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
The origin (or O-world) is the basic polyhedron 1∗ , the 4-valent graph with
one vertex, a usual symbol of infinity (∞) (Fig. 1.43c). The first, linear
world (or L-world) contains only one source link: the Hopf link 2 (221 in the
classical notation). We use it to derive the family p of alternating KLs,
with KL shadows represented by p-gons with bigonal edges (p ≥ 2). For
odd p we have the infinite series of alternating knots 3, 5, 7, . . . (or 31 , 51 ,
71 , . . .), and for even p the infinite series of alternating 2-component links
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
70 LinKnot
2, 4, 6, . . . (or 221 , 421 , 621 , . . .) (Fig. 1.47). For n > 1 the members of the
linear world can be included in the next, rational world.
Rational world (or R-world) consists of rational links.
From this definition we can compute the number of rational KLs with
n crossings.
Proof. Think of n as a linearly ordered set of, say, stars; then choosing a
composition amounts to choosing a subset of the set of n−1 spaces between
the stars. For example,
∗∗ | ∗ | ∗ ∗ ∗ | ∗
(choices of spaces indicated by bars) is the composition 2 1 3 1 of 7.
Suppose that n ≥ 3, and let bn denote the number of compositions of
n with no 1 in either first or last position, and where a composition is
identified with its reverse.
Compositions of n not having 1 as either first or last part correspond
to sets of spaces between n ordered dots not containing either the first or
last space, so there are an = 2n−3 of these.
If sn of these are symmetric (i.e. equal to their reverses) then we have
an − sn an + sn
bn = + sn = .
2 2
Choosing a symmetric composition of n without 1 in first or last place
corresponds to choosing a subset of the set of spaces up to and including
the middle space (if there is one) but excluding the first space. (The rest
of the spaces are determined by symmetry). There are [ n−2 2 ] such spaces,
[ n−2 ]
and thus sn = 2 2 .
Hence
n−2
2n−3 + 2[ 2 ] n−2 n
bn = = 2n−4 + 2[ 2 −1] = 2n−4 + 2[ 2 ]−2 .
2
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
This simple formula was first derived by C. Ernst and D.W. Sumn-
ers (1987) in another form, and later independently by S. Jablan (1999a,
1999b). For n ≥ 4 we can compute the first 20 numbers of this sequence.
For n = 3 we have one knot, so the sequence is: 1, 2, 3, 6, 10, 20, 36,
72, 136, 272, 528, 1056, 2080, 4160, 8256, 16512, 32896, 65792, 131328,
262656, 524800, . . . This sequence is included in On-Line Encyclopedia of
Integer Sequences (http://www.research.att.com/∼njas/sequences/) as the
sequence A005418. The number of rational knots with n crossings (n ≥ 3)
is given by the formula
n (n−1) (mod 2) n
2n−3 + 2[ 2 ]−2 + (−1)(n−1)[ 2 ] (mod 2)
3
(giving the sequence A090596), so we can derive the formula for the number
of rational links with n crossings as well.
The LinKnot function RationalKL calculates the number and Conway
symbols of all rational KLs for a given number of crossings n. The results
are given in the following tables:
n=9
24 Knots
9 72 63 54 522 513
423 4212 4122 41112 342 333
3222 3213 31212 31122 311112 2412
2322 23112 22122 21312 212112 2111112
12 Links
612 5112 432 414 4113 3312
32112 3132 31113 252 22212 221112
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
72 LinKnot
n = 10
45 Knots
82 712 64 613 6112
532 5212 514 5113 51112
433 4312 4222 42112 4132
4123 4114 41122 41113 352
3412 3313 33112 3232 32212
32113 321112 31312 31222 312112
31132 311122 311113 2512 2422
24112 2332 23122 22312 222112
221212 2211112 212212 2121112 21111112
27 Links
73 622 55 52 3 5122
442 424 4213 412 12 411112
343 3322 3223 321 22 3142
31213 311212 3111112 26 2 23212
231112 22222 221122 214 12 213112
2112112 10
For n ≥ 4 we obtain the sequence 1, 1, 2, 2, 4, 5, 9, 12, 21, 30, 51, 76, 127,
195, 322, 504, 826, 1309, 2135, 3410 . . ., known as the sequence A102526.
Both of these sequences, A005418 and A001224, have been discovered be-
fore, but in a different context, related to “Binary grids” and “Packing a
box with n dominoes”. Rational source KLs are given in the following
table:
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
n No. KLs
2 1 2
4 1 22
5 2 212
6 2 222 2112
7 2 2212 21112
8 4 2222 21212 22112 211112
9 5 2212 2 22212 212112 221112
2111 112
10 9 2222 2 212212 221122 221212
2221 12 2112112 2121112 2211112
2111 1112
n No. KLs
2 1 2
3 1 3
4 1 2 2
5 2 3 2 2 1 2
6 4 3 3 2 2 2 312 2112
7 6 2 3 2 3 1 3 322 2212 3112
8 11 2 1 1 12 3 2 3 332 2222 2312
3 1 1 3
9 18 3 3 3 2 3 2 2 3 1 3 2 3213 3222
3 3 1 2 2 1 3 12 2 2 1 22 22212 23112
3 1 1 13 3 1 1 22 3 1 2 12 32112 212112
2 2 1 112 3 1 1 112 2 1 1 1112
n = 10
2332 3223 3232 3313 3322
22222 22312 23122 23212 31132
31213 31222 31312 32113 32122
32212 33112 213112 212212 221212
221122 222112 231112 311113 311122
311212 312112 321112 2112112 2121112
2211112 3111112 21111112
No. of KLs: 33
Results in the tables above were computed with the LinKnot function
RatGenSourKL which gives the number and Conway symbols of all ra-
tional source KLs and rational generating KLs with n crossings. We can
recognize a regular distribution of the source links and their corresponding
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
74 LinKnot
Fig. 1.54 Shadows of source links (where exponent l denotes l-fold repetition).
The first solution for rational KLs is given below: take a rational link in
Conway notation and reduce all numbers mod 2. Then apply the following
cancellation rules:
(1) for every sequence of the form xa0 (a ∈ {0, 1}), xa0 = x;
(2) for every sequence of the form xa1 (a ∈ {0, 1}), a1 = 1 − a.
76 LinKnot
Theorem 1.18. Two rational tangles are equivalent iff their continued
fractions yield the same rational number (Conway, 1970).
′
Unoriented rational links L( pq ) and L( pq′ ) are ambient isotopic iff:
(1) p = p′ and
(2) either q ≡ q ′ (mod p) or qq ′ ≡ 1 (mod p) (Schubert, 1956).
The proof of the Theorem 1.18 can be found in J.M. Montesinos (1984),
G. Burde and H. Zieschang (1985), or in the papers by J.R. Goldman
and L. Kauffman (1997), and the direct combinatorial proof is given by
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
For example, Hopf link 222 (2) is achiral as non-oriented and chiral as an
oriented link.
78 LinKnot
For the knot 2 2, the graph symmetry group is G = [2+, 4], and the knot
symmetry group G′ = [2+, 4+] is generated by the rotational reflection,
with the axis defined by the midpoints of colored (i.e., double) edges of the
tetrahedron (Fig. 1.56). If we take into account the signs of crossings, it is
a rotational antireflection. Its effect is preserved in all rational knots with
an even number of crossings which have a mirror-symmetric (palindromic)
Conway symbol.
Theorem 1.20. A rational knot is achiral iff its Conway symbol is mirror-
symmetric and has an even number of crossings (Siebenmann, 1975; Cau-
dron, 1982).
n=4
22
n=6
2112
n=8
44 3113 2222
n = 10
4114 311113 2332 212212 21111112
n = 12
66 5115 4224 3333 2442
321123 312213 222222 22111122 21211212
2111111112
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
Theorem 1.21. All oriented rational KLs are invertible (Kauffman and
Lambropoulou, 2002a).
According to this, a knot can have several different periods. For exam-
ple, a trefoil knot has periods 2 and 3. The period of the knot 2 2 is 2, and
the same holds for all rational achiral knots. If the antisymmetry group of
the vertex-bicolored graph corresponding to an achiral oriented projection
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
80 LinKnot
n=2
2
n=6
33
n=8
211112
n = 10
55 3223 221122
n = 12
411114 31111113 231132 213312
21122112
n = 14
77 5225 421124 3443
322223 31211213 241142 223322
22122122 2121111212 211111111112
Theorem 1.22. A rational non-oriented link is achiral iff its Conway sym-
bol is mirror-symmetric (palindromic) and has an even number of crossings
(Kauffman and Lambropoulou, 2002).
Fig. 1.58 A crossing change (a) and the Nakanishi-Bleiler example: (b) the minimal
projection of the knot 5 1 4 that requires at least three crossing changes to be unknotted;
(c) the minimal projection of the knot 3 1 2 with the unknotting number 1; (d) non-
minimal projection of the knot 5 1 4 from which we obtain the correct unknotting number
u(5 1 4) = 2.
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
82 LinKnot
The first estimation of the unlinking number follows from the theorem:
Corollary 1.2. For an n-crossing link L and its unlinking number u(L)
the following inequality holds: u(L) ≤ [ n2 ].
The estimation given in this corollary is a very rough, but still useful.
Difficult knots are extremely rare, and the question of finding all difficult
KLs with a given number of crossings n is open.
(1) according to the classical definition, one is allowed to make a planar iso-
topy after each crossing change and then continue the unlinking process
with the newly obtained projection, until an unlink is obtained;
(2) the standard definition requires all crossing changes to be done simul-
taneously in a fixed projection.
On the other hand, making a crossing change in the middle point of the
diagram (Fig. 1.58b) followed by the reduction 5 −1 4 = 3 1 2, we obtain the
minimal projection of the knot 3 1 2 (Fig. 1.58c) that can be unknotted by
one crossing change. Hence, in this case unknotting according the classical
definition, using only minimal projections, gives correct unknotting number
2. The unknotting number can also be obtained from the non-minimal
projection of the knot 5 1 4 (Fig. 1.58d) using the standard definition. As
it was shown by J. Bernhard (1994), the same property holds for the whole
knot family (2k + 1) 1 (2k), k ≥ 2.
Nakanishi-Bleiler example motivated us to define BJ-unlinking number
which is computable due to the algorithmic nature of its definition.
84 LinKnot
This means that we take all minimal projections of a KL, make a cross-
ing change in every crossing, and then minimize all the projections obtained.
The same algorithm is applied to the first, second, . . . k th generation of the
KLs obtained. The BJ-unlinking number is the minimal number of steps
k in this recursive unlinking process.
If BJ-conjecture does not hold for all KLs, it may be true for some
restricted class, e.g., for all rational KLs. Notice that, even if BJ-conjecture
does not hold for all links, BJ-unlinking number is the best computable
upper bound for the unlinking number, since δBJ (L) ≤ δ(L).
Going further in the same direction, we conjecture that unlinking num-
ber is a linear subfamily-dependent KL invariant (Conjecture 1.2).
We illustrate the importance of the conjecture by the following exam-
ple. Consider alternating pretzel knots a, b, c where 1 < a ≤ b ≤ c and
a, b, c are all odd numbers. We show in the following proposition that
uBJ (a, b, c) = a+b
2 . However, the unknotting number of these knots is still
unknown, except for the smallest knots, e.g., 3, 3, 3 with unknotting number
3, computed recently by B. Owens (2005).
Proposition 1.1.
a) for rational knots with the Conway symbol (2m+ 1) 1 (2n+ 1), (m ≥ n)
BJ-unknotting number is uBJ = n + 1;
b) for pretzel knots a, b, c (1 < a ≤ b ≤ c)4 , where a, b, c are odd numbers:
a+b
uBJ (a, b, c) =
2
For mutant KLs, the following lemma holds:
4 14 18 12 28 26 − 22 2 − 24 10 8 − 16 6 20
and
6 − 10 26 20 − 2 − 18 − 22 28 − 24 8 − 14 − 12 4 16
(Fig. 1.59), give two different BJ-unknotting numbers, 1 and 2, respec-
tively.
86 LinKnot
Theorem 1.24. All rational KLs with unlinking number one have an un-
linking number one minimal diagram.
Theorem 1.25. For the distance of knots K1 and K2 , the following in-
equality holds: d(K1 , K2 ) ≥ |σ(K1 )−σ(K
2
2 )|
(Murakami, 1985; Kawauchi,
1996).
88 LinKnot
(1, 11 , 1)
(2, 31 , 3)
(3, 41 , 2 2)
(4, 51 , 5) (5, 52 , 3 2)
G1 = {{2, 1}, {3, 1}, {4, 2}, {5, 1}, {5, 2}, {5, 4}, {6, 1}, {6, 3}, {7, 1}, {7, 2}, {7, 3},
{7, 6}, {8, 1}, {8, 2}, {9, 4}, {10, 1}, {10, 5}, {11, 4}, {11, 5}, {11, 9}, {11, 10}, {12, 2},
{12, 5}, {12, 11}, {13, 2}, {13, 4}, {13, 5}, {13, 9}, {13, 10}, {14, 1}, {14, 2}, {14, 3}, {14, 5},
{14, 10}, {15, 1}, {15, 2}, {15, 3}, {16, 1}, {16, 6}, {17, 2}, {17, 4}, {17, 7}, {18, 6}, {18, 16},
{19, 3}, {19, 6}, {19, 7}, {19, 18}, {20, 7}, {21, 2}, {21, 6}, {21, 7}, {21, 16}, {21, 17},
{21, 19}, {21, 20}, {22, 1}, {22, 2}, {22, 4}, {22, 8}, {23, 2}, {23, 5}, {23, 8}, {23, 22},
{24, 1}, {24, 7}, {24, 19}, {25, 4}, {25, 5}, {25, 8}, {25, 23}, {26, 1}, {26, 5}, {26, 6}, {26, 7},
{26, 16}, {26, 17}, {27, 3}, {27, 6}, {27, 16}, {28, 1}, {28, 3}, {28, 5}, {28, 8}, {28, 22},
{29, 1}, {29, 2}, {29, 3}, {29, 5}, {29, 7}, {29, 17}, {30, 4}, {30, 5}, {31, 2}, {31, 3}, {31, 5},
{31, 8}, {32, 1}, {32, 2}, {32, 7}, {33, 2}, {34, 4}, {34, 13}, {34, 30}, {35, 1}, {35, 5}, {35, 8},
{35, 14}, {35, 25}, {36, 1}, {36, 4}, {36, 7}, {36, 20}, {37, 9}, {38, 1}, {38, 10}, {39, 9},
{39, 11}, {40, 10}, {40, 11}, {41, 5}, {41, 10}, {41, 12}, {42, 4}, {42, 9}, {42, 13}, {43, 2},
{43, 10}, {43, 13}, {44, 2}, {44, 3}, {44, 6}, {44, 14}, {45, 9}, {45, 11}, {45, 13}, {46, 11},
{46, 12}, {47, 4}, {47, 5}, {47, 11}, {47, 14}, {48, 1}, {48, 6}, {48, 10}, {48, 14}, {49, 4},
{49, 11}, {49, 12}, {49, 13}, {50, 1}, {50, 2}, {50, 5}, {50, 6}, {50, 15}, {51, 3}, {51, 5},
{51, 10}, {51, 14}, {52, 13}, {52, 34}, {53, 3}, {53, 7}, {53, 15}, {54, 5}, {54, 10}, {54, 11},
{54, 13}, {55, 1}, {55, 3}, {55, 6}, {55, 15}, {56, 2}, {56, 5}, {56, 7}, {56, 13}, {56, 14},
{57, 1}, {57, 2}, {57, 5}, {57, 12}, {57, 14}, {58, 1}, {58, 7}, {58, 8}, {58, 15}, {59, 4}, {59, 5},
{59, 13}, {60, 1}, {60, 7}, {60, 14}, {60, 35}, {61, 5}, {61, 6}, {61, 7}, {61, 14}, {62, 1},
{62, 2}, {62, 4}, {62, 5}, {62, 7}, {62, 15}, {63, 1}, {63, 2}, {63, 3}, {63, 7}, {63, 8}, {63, 14},
{64, 1}, {64, 8}, {64, 36}, {65, 8}, {65, 14}, {65, 15}, {66, 1}, {66, 6}, {66, 7}, {66, 8}, {67, 2},
{67, 4}, {67, 5}, {67, 8}, {68, 2}, {68, 3}, {68, 5}, {68, 6}, {68, 7}, {68, 15}, {68, 36}, {69, 1},
{69, 2}, {69, 3}, {69, 8}, {69, 14}, {69, 35}, {70, 1}, {70, 2}, {70, 3}, {70, 7}, {70, 15},
{71, 12}, {72, 4}, {72, 13}, {72, 14}, {73, 6}, {73, 15}, {74, 4}, {74, 12}, {74, 13}, {75, 1},
{75, 7}, {75, 12}, {75, 14}, {76, 3}, {76, 36}, {77, 5}, {77, 15}, {78, 1}, {78, 6}, {78, 7},
{78, 15}, {78, 21}, {78, 27}, {78, 32}, {79, 4}, {79, 5}, {79, 7}, {79, 14}, {79, 29}, {80, 1},
{80, 6}, {80, 8}, {80, 14}, {80, 15}, {80, 23}, {80, 31}, {81, 1}, {81, 4}, {81, 5}, {81, 7},
{81, 13}, {81, 14}, {82, 2}, {82, 6}, {82, 15}, {82, 26}, {82, 36}, {83, 2}, {83, 6}, {83, 15},
{84, 2}, {84, 12}, {84, 35}, {85, 4}, {85, 12}}.
90 LinKnot
“left” and “right” form of the same knot. Examples of alternating knots
with n crossings (n ≤ 8) and the Gordian distance 1 achieved from the
non-minimal diagrams with n + 1 crossings are:
n=5 (5),(3 2)
n=6 (4 2),(3 1 2)
n=7 (7),(4 3) (7),(3 2 2)
(5 2),(4 3) (5 2),(3 2 2)
(5 2),(2 2 1 2) (4 3),(3 1 3)
(2 3 1 2), (2 1, 3, 2)
n=8 (3, 3, 2), (2 1, 2 1, −2) (3, 3, −2), (2 1, 2 1, 2)
(3, 2 1, −2), (2 1, 3, 2) (5 1 2), (2 2 1 1 2)
(6 2), (2 2 2 2) (4 1 1 2), (2 3 1 2)
(4 1 1 2), (3 1 1 1 2) (4 1 3), (3 1 1 3)
(5 1 2), (3 2 1 2) (6 2), (3 2 1 2)
(3 3 2), (3, 3, 2) (4 1 3), (3 3 2)
(5 1 2), (3 3 2) (6 2), (3 3 2)
(4 4), (4 1 3) (6 2), (4 4)
Knots 4 2 (61 ) and 2 3 1 2 (88 ) are examples of knots with the distance
1 from their mirror images (Fig. 1.62).
The LinKnot function AllStatesRational uses a Conway symbol of a
rational KL projection P as an input and makes all possible combinations of
crossing changes on P , i.e., choices of signs ±. The output is the unlinking
number of the fixed projection P , followed by a list of ordered pairs where
each second entry is a rational knot or link L, and the first entry the
minimal number of crossing changes required to obtain L from P . The
function AllStatesRational is very fast, since it is based on continued
fractions. A similar, but much slower function, based on the function Re
ductionKnotLink, can be applied to all KLs. It gives the same output,
where knots or links L are given by their reduced P -data.
The functions AllStatesRational and UnR can be used to illustrate
the Nakanishi-Bleiler example (knot 5 1 4), or similar examples of knots
from the family (2k + 1) 1 (2k), showing that the unknotting number of
the fixed minimal projection of a knot (2k + 1) 1 (2k) is k + 1, and its
BJ-unknotting number is k, so the BJ-unknotting gap is δBJ = 1. We
can also consider links from the family (2k) 1 (2k) (k > 1). A fixed minimal
projection of a link (2k) 1 (2k) can be unlinked with minimum 2k−1 crossing
changes, and BJ-unlinking number, computed by the function UnR, is k,
so the BJ-unlinking gap is k − 1.
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
92 LinKnot
Fig. 1.62 (a) Transition from 4 2 and (b) from 2 3 1 2 to its mirror image by one crossing
change.
However, in that case our list may not be complete: KLs whose un-
linking number is smaller than the BJ-unlinking number will also have an
unlinking gap.
The following tables contain the Conway symbols of rational knots and
links up to 16 crossings with a non-trivial BJ-unlinking gap, given accord-
ing to the number of crossings and whether they are knots or links. Symbols
given in bold denote the links with the BJ-unlinking gap 2 (the others have
BJ-unlinking gap 1). The first column in each table gives the number of
crossings, second the number of KLs with non-trivial BJ-unlinking gap,
and the third column their list.
n = 13
7 Knots
4414 6142 41314 51322 231412
511132 513112
16 Links
6 1 6 634 814 5152 5332
6133 7132 34132 41422 51115
61123 71113 241312 411142 611122
4211113
n = 14
31 Knots
716 734 914 4163 4343
5414 6143 6152 6332 7133
8132 33152 35132 41423 51314
51323 51422 61124 61322 71123
81113 313142 314132 341312 351113
511142 711122 2141312 2411132 4211114
5211113
5 Links
41432 41612 413132 513212 5131112
Among the links with n = 13 crossings we find the first link 6 1 6 with
the BJ-unlinking gap δBJ = 2.
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
94 LinKnot
n = 15
43 Knots
21314112 2241312 24111312 2411142
241422 321162 331512 33162
341322 4111422 4141113 414132
414222 41433 41514 41613
41712 42111132 4434 451113
45132 4614 5111214 511134
513114 513213 51324 51432
51522 51612 53322 61112112
611232 6131112 61314 613212
61332 6144 6162 6414
711132 713112 8142
63 Links
10 1 4 21413112 23111412 2 31412 2
241512 25111212 251322 3 11152 2
341313 34152 34332 3 51112 2
35133 36132 41111232 41 21111 13
4141212 41442 41622 4 21131 3
421152 4221123 4231113 42414
4311213 43422 441312 511112 4
5111223 5111322 511143 51315
51513 52111122 5211123 5352
54132 5532 611142 61125
61224 61323 61422 6153
6211113 63123 6333 6 36
711123 7 1 1 1 5 711222 71133
71313 7134 7 15 2 7332
811122 81123 8133 8 16
834 91113 9132
n = 16
138 Knots
10 1 1 1 3 10 1 3 2 11 1 4 214111312
214131112 215111212 2231512 231622
232111132 241432 241612 2511232
2611222 31113142 31211152 3124132
31311142 3131422 31411132 31413112
314332 31511122 3214132 323152
3241312 3251113 33111412 331513
33352 334132 3411142 341422
341512 34211113 35111212 351322
35152 35332 3611122 361123
36133 371113 37132 4111423
412111114 413143 4141213 414142
4142122 414223 414232 414322
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
First rational knots with the non-trivial unknotting gap δBJ = 2 are
6 1 6 3 and 8 1 5 2 with n = 16 crossings. First non-rational alternating
knots with the BJ-unlinking gap δBJ = 1 appear for n ≥ 12 crossings:
pretzel knot 5, 4, 3 (12a1242 ) and polyhedral knots 6∗ 2.4 0 : 3 0 (12a970 ),
6∗ 2.2 1 0 : 4 0 (12a76 ), and 6∗ 2.2.2.4 0 (12a1153 ).
We will try to explore the effect which 2n-moves (Przytycki, 2006a) (Fig.
1.67a) have on the BJ-unlinking number uBJ , diagram unlinking number
uM , and BJ-unlinking gap δBJ . Applying 2n-moves on an integer tangle
decreases or increases its corresponding parameter in the Conway symbol
for 2n. If we allow applying 2n-moves on an arbitrary subset of elementary
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
96 LinKnot
Lemma 1.2. For a rational link with the Conway symbol a b, the following
holds:
• if a, b are both odd, then for a link (2m + 1) (2n + 1) we have
uBJ = uM = u = m + n + 1;
• if a is odd and b is even, then for a knot (2m + 1) (2n) we have
uBJ = uM = u = n;
• if a, b are both even, then for a knot (2m) (2n), (m ≥ n) we have
uBJ = uM = n.
Lemma 1.3. For a rational link with the Conway symbol a b c, the follow-
ing holds:
i.e.
uBJ = min(k + m, k + l)
and
k + m, if 1 ≤ l ≤ k, m;
uM = k + m − 1, if 0 = l ≤ k, m;
k + m, if k, m ≤ l.
Among the KL families from Lemma 1.3, two of them have non-trivial
BJ-unlinking gap:
Corollary 1.3.
• rational links (2k) (2l + 1) (2m) have non-trivial BJ-gap if k ≥ m ≥ 2
and m ≥ l + 1
m − 1, l = 0;
δBJ =
m − l, l ≥ 1.
• rational knots (2k + 1) (2l + 1) (2m) have non-trivial gap δBJ = 1 if
m ≥ 2 and l + 1 < m < k.
Lemma 1.4. For a rational knot with the Conway symbol (2k) (2m) 1 (2n)
the following holds:
• diagram unlinking number is
uM = n + min(k, m − 1);
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
98 LinKnot
• BJ-unlinking number is
n, if m ≤ n;
uBJ =
n + min(k, m − n), if m > n.
• BJ-unlinking gap is
min(k, m − 1), if m ≤ n;
δBJ =
min(k, m − 1) − min(k, m − n), if m > n.
The Lemma 1.2 is used to prove the following theorem about a family
of rational knots with an arbitrarily large BJ-unlinking gap6 :
The main problem for every family is finding necessary and sufficient
conditions (or, simply, conditions) for a specific family to have a non-trivial
unlinking gap.
Knot or link a is a torus KL of type [2, a] and therefore uBJ = uM = u,
so both gap and BJ-unlinking gap are trivial. From Lemma 1.2 it follows
that all rational links with 2 parameters a b have trivial BJ-unlinking gap.
The same holds for all 3-parameter families a b c except two families listed
in Corollary 1.3. Since computations, based on parity of parameters and
symmetries of the links are long and tedious even for 3-parameter families,
for multi-parameter families we give only experimental results.
We choose 68 one-parameter7 families of rational links which contain all
rational links up to 14 crossings with positive BJ-unlinking gap8 . For all
6 S. Bleiler (1984) asked if δ(L) = u (L) − u(L) has an upper bound. Since δ
M BJ (L) ≤
δ(L), Theorem 1.26 provides more examples of links with unbounded δ(L) (compare with
Stoimenow, 2003).
7 One-parameter family is obtained by applying the same 2n-move to all chosen integral
tangles.
8 Compare with the tables, pages 93-95.
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
100 LinKnot
• the family (2k) (2m) 1 (2n) has an arbitrarily large BJ-unlinking gap
(see Theorem 1.26);
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
• the family (2k + 2) 1 (2k + 2) (2k − 1), (k ≥ 1), starting with the knot
6 1 6 3 has uBJ = k + 1 and δBJ = k;
• the family (4k +4) 1 (2k +3) (2k), (k ≥ 1), starting with the knot 8 1 5 2,
has uBJ = k + 2 and δBJ = k + 1;
• the family (2k + 1) (2l − 1) (2m − 1) (2n − 1) (2p + 1), starting with
the link 5 1 1 1 3, has a subfamily (2k + 1) 1 1 1 (2l + 1) which under the
conditions k ≥ l ≥ 0 has unlinking gap
0, if k = l = 0;
δ = δBJ = l − 1, if k = l > 0;
l, if k > l > 1.
(k1 , . . . , k2i+1 ≥ 2, l < mini (ki )) and every link family of the form
Next we consider the family of pretzel knots a, b, c. For the pretzel knots
2k + 1, 2l + 1, 2m + 1 (k ≥ l ≥ m ≥ 1) we proved (Proposition 1.1b) that
uBJ = l + m and δBJ = 0. For the simplicity, since pretzel KLs with three
tangles remain unchanged by a permutation of tangles, we will assume that
in their symbols tangles are given in non-increasing order. For the families
of pretzel KLs with three tangles we have the following:
Theorem 1.27.
9 Notice that the linking number guarantees that uBJ = uM = u and δ = δBJ .
10 Notice that in first two cases, l = 1 and k ≥ l > 1, the signature guarantees that
uBJ = uM = u and δ = δBJ .
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
102 LinKnot
2 2 1 1 2, 2, 2 4 1 1, 3, 3 2 1 1, 3 1, 3 1 5, 4, 3
3 1, 3 1, 2 1+ (2 1, 2 1 1 1) (2, 2) (2 1, 2 2) 1 (2, 2)
104 LinKnot
numbers. Two different minimal projections of the knot 1436750 (Fig. 1.59)
can be extended to the two-parameter families of minimal projections
124∗ − 1. − 1. − 1. − 1. − 1 : (−1, (−2l)) 0 : (2k − 1) : . − 1
and
8∗ − 2 0 : (2l + 1) 0 : −2 0.(−2k) 0. − 1.2 0
which satisfy the condition l < 2k (Fig. 1.64), where 124∗ denotes the basic
polyhedron 12D from the paper by A. Caudron (1982). The experimental
BJ-unknotting number of the first family of projections is k, and it is the
experimental BJ-unknotting number of the knot family in question. The
second family of projections has the experimental BJ-unknotting number
k + 1.
It is not surprising that some non-minimal diagrams can have a non-
trivial unlinking gap. For example, the 11-crossing non-alternating knot
11n138 has the non-minimal diagram 3 1 1, 3, 3− with the unknotting gap
δM = 1, while the (fixed) minimal diagram 3 1 1, 3, −2 1 gives the unknot-
ting number u = 2 (Fig. 1.65).
The family of non-alternating pretzel links 2k, 2k, −3 (k ≥ 2) is the
candidate for non-alternating link family with an arbitrarily large unlinking
gap (Fig. 1.66).
This family is obtained from the family of rational links 2k 1 2k =
2k, 2k, 1 (k ≥ 2) which is a special case of the family 2k, 2l + 1, 2m (from
the Corollary 1.3 for l = 0, k = m), with arbitrarily large BJ-unlinking gap
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
Fig. 1.65 (a) Non-minimal diagram 3 1 1, 3, 3−; (b) the minimal diagram 3 1 1, 3, −2 1
of the non-alternating knot 11n138 .
δBJ = k−1. In the similar manner as before, we may obtain that the family
of standard minimal diagrams of pretzel links 2k, 2k, −3 has BJ-unlinking
number k. Furthermore, the unlinking number of the standard diagram of
2k, 2k, −3 is equal to 2k − 1, hence the diagram BJ-unlinking gap is k − 1.
Since the classification of all minimal diagrams of the link family
2k, 2k, −3 is, up to our knowledge, not yet achieved we are not able to
show that the link family 2k, 2k, −3 has an arbitrarily large unlinking gap.
One-parameter family of minimal link diagrams 2k, 2k, −3 can be ex-
tended to a three-parameter family 2k, 2l, −(2m + 1) with an arbitrarily
large diagram BJ-unlinking gap.
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
106 LinKnot
Fig. 1.66 The family 2k, 2k, −3 of non alternating minimal diagrams with an arbitrarily
large BJ-unlinking gap.
Next we will consider two special classes of rational KLs: KLs with
unknotting (unlinking) number equal to 0 or 1. For rational KLs with un-
knotting (unlinking) number 1, the unknotting (unlinking) number can be
recognized from any minimal projection (Theorem 1.24; Nakanishi, 1996;
Stoimenow, 2004). Every unknotting number one knot is prime (Scharel-
mann, 1985). The general form of knots with unknotting number 1 (The-
orem 1.28) is described by T. Kanenobu and H. Murakami (1986), and
P. Kohn formulated an analogous theorem for links (1991) (Theorem 1.29).
Theorem 1.24 is the corollary of Theorem 1.28 and 1.29, because a single
crossing change in minimal diagrams from Theorem 1.28 and 1.29 gives the
unknot and unlink.
Moreover, if L is a rational link with a minimal diagram which can
not be unlinked by one crossing change and has a minimal diagram with
unlinking number 2, then u(L) = 2.
The LinKnot function RatKnotGenU1 gives the number and the list
of Conway symbols of all rational knots with the unknotting number 1 with
n crossings, and the function RatLinkU1 gives the same result for rational
links with the unlinking number 1. The number of such knots is given by
the formula
n−2
2[ 2 ] − 1.
The number of rational unlinking number 1 links is 0 for every even n, and
for odd n it is given by the formula
n−7
2[ 2 ] .
In a similar way we can be interested in rational representations of
unknots or unlinks, i.e., KLs with the unknotting (unlinking) number 0.
Theorem 1.30. Every rational unknot can be expressed by one of the fol-
lowing Conway symbols
c0 c1 . . . cr−1 cr (−1) 1 (cr − 1) cr−1 . . . c1
Theorem 1.31. Every rational unlink can be expressed by one of the fol-
lowing Conway symbols
c0 c1 . . . cr−1 cr (−1) 1 (cr − 1) cr−1 . . . c1 c0
Proof of these two theorems follows from the proof of Theorem 1.28
(Kanenobu and Murakami, 1986) and Theorem 1.29 (Kohn, 1991).
The LinKnot function RatKnotGenU0 produce, for a given n, a list of
Conway symbols of all rational unknots with n crossings, and the function
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
108 LinKnot
RatLinkU0 gives the same result for rational unlinks. The number of
rational unknots with n crossings is given by the formula
n−2
2[ 2 ] − 2.
The number of rational unlinks is 0 for every even n, and for odd n it is
given by the formula
n−7
2[ 2 ] .
For example, knot illustrated in Figure 1.67 has the unknotting number
1. Try to find the crossing change that results in unknot. If you don’t be-
lieve that such crossing change exists, you can check its unknotting number
by entering its Conway symbol
321123311232112
in the LinKnot function UnR. In a similar way you can check that the link
3211233 −112321123
from Fig. 1.68 is an unlink.
Another way of checking if a KL is an unknot or unlink is exporting it
to KnotPlot, using the function fCreateGraphics and relax it, so one can
see how a complicated projection reduces to one or several disjoint circles.
unknot.
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
110 LinKnot
Fig. 1.70 (a) n-move; (b) link which can not be reduced to a trivial link by 3-moves;
(c) knot 16101∗ − 1. − 1 : . − 1. − 1 : −1. − 1 :: . − 1 : −2 0.
1315∗2 1 0. − 1. − 1. − 1. − 1 : .2 0.1.3 0. − 1. − 1. − 1
By two 4-moves, it gives the knot
1315∗ − 2 1 0. − 1. − 1. − 1. − 1 : .2 0.1. − 1 0. − 1. − 1. − 1
which reduces to 12-crossing knot 2 0.− 2 − 1.− 2 0.2 1 0. From it, we obtain
the unknot by the following sequence of 4-moves and ambient isotopies
16101∗ − 1. − 1 : . − 1. − 1 : −1. − 1 :: . − 1 : −2 0
(2, 1)-cable of the figure-eight knot with n = 17 crossings proposed by
N. Askitas (Fig. 1.70c) (Askitas, 1999; Przytycki, 2002, 2003).
An unlinking operation can be defined:
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
112 LinKnot
Fig. 1.71 (a) (2, 2)-move and (−2, −2)-move; (b) (2, 2)-move 1 → (−2, −1).
114 LinKnot
Fig. 1.72 (a) Knots 9∗ (940 ) and −2 0 : −2 0 : −2 0 (949 ); (b) links 2 : 2 : 2 and
2 : −2 0 : −2 0.
Fig. 1.73 (a) Knots 3 0 : 2 : 2 and 3 0 : −2 0 : −2 0 that reduce to the mirror image of
949 ; (b) knots −3 : 2 : 2 and −3 : −2 0 : 2 that reduce to the mirror image of 949 .
n = 11 1 11a297 9∗ 2 0 :::: 2 0 →
9∗ 2 0 ::: . − 2 − 1 0. − 2 ≃ 9∗ 940
n = 11 1 11a317 6∗ 3 1.2.2 0 →
6∗ 3 1. − 2 0. − 2 ≃ −3 : 2 : 2 →
−3 : −2 0 : 2 ≃ −2 0 : −2 0 : −2 0 949
n = 11 1 11n133 8∗ 2. − 2 0.2 →
8 − 2 0. − 2 0.2 ≃ 9∗
∗ 940
n = 11 1 11n148 9∗ . − 2 − 1 →
9∗ .1 ≃ 9∗ 940
n = 11 2 6∗ 3.2.3 →
6∗ (−2, 1). − 2 0.3 ≃ −3 : 2 : 2 →
−3 : −2 0 : 2 ≃ −2 0 : −2 0 : −2 0 949
n = 11 2 8∗ 2.2.2 →
8∗ − 2 0. − 2 0.2 ≃ 9∗ 940
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
116 LinKnot
n = 12 1 12a100 6∗ (2 1, 2).2 0 :: 2 0 →
6∗ (−1, 2).2 0 :: 2 0 ≃ −2 0 : −2 0 : −2 0 949
n = 12 1 12a780 6∗ 2.2.2 1.2 0.2 0 →
6∗ − 2 0.2. − 1.2 0.2 0 ≃ −2 0 : −2 0 : −2 0 949
n = 12 1 12a907 9∗ 3 :::: 2 0 →
9 3. − 2 − 1 0 ::: . − 2 ≃ 9∗ . − 2 − 1 →
∗
9∗ .1 ≃ 9∗ 940
n = 12 1 12a921 6∗ 3.2.3 1 →
6∗ 3. − 2 0.3 1 ≃ −2 1 0.3.2.2 0 →
1.3. − 2 0.2 0 ≃ −3 : 2 : 2 →
−3 : −2 0 : 2 ≃ −2 0 : −2 0 : −2 0 949
n = 12 1 12a975 8∗ 2 0 : 2 0 : 2 0 : 2 0 →
8∗ − 2 : −2 : 2 0 : 2 0 ≃ −3 : 2 : 2 →
−3 : −2 0 : 2 ≃ −2 0 : −2 0 : −2 0 949
n = 12 1 12a1194 8∗ 2.3 0.2 →
8∗ − 2 0.(−2, 1) 0.2 :: . − 2 − 1 0 ≃ 9∗ 2 0 :::: 2 0 →
9∗ 2 0 ::: . − 2 − 1 0. − 2 ≃ 9∗ 940
n = 12 2 6∗ 3 1 1.2.2 0 →
6∗ 3 1 1. − 2 0.2 0 ≃ 3.2.2 0. − 2 1 0 →
3. − 2 0.2 0.1 ≃ −3 : 2 : 2 →
−3 : −2 0 : 2 ≃ −2 0 : −2 0 : −2 0 949
n = 12 2 6∗ (2, 2+).2 0 :: 2 0 →
6∗ 2 − 1.2 0 :: 2 0 ≃ −2 0 : −2 0 : −2 0 949
n = 12 2 6∗ (2, 2) 1.2 0 :: 2 0 →
6∗ 2 − 2 1.2 0 :: 2 0 ≃ 3 0 : 2 : 2 →
3 0 : −2 0 : −2 0 ≃ −2 0 : −2 0 : −2 0 949
n = 12 2 6∗ 2.2.2.2 0.2 0.2 0 →
6∗ 2.2. − 2 0.2 0.2 0.2 0 ≃ 9∗ . − 4 0 →
9∗ .2 − 2 0.(−2, −1) ≃ 9∗ 940
n = 12 2 8∗ 2.2 0.2 : .2 →
8 − 2 0. − 2.2 : .2. − 2 − 1 ≃ 9∗
∗ 940
n = 12 2 9∗ (2, 2) →
9 2 − 2 ≃ 8∗ 2. − 2 0.2 →
∗
8∗ − 2 0. − 2 0.2 ≃ 9∗ 940
n = 12 2 9∗ .(2, 2) →
9∗ .2 − 2 ≃ 9∗ . − 2 − 1 →
9∗ .1 ≃ 9∗ 940
n = 12 3 6∗ 2.(2, 2).2 : 2 →
6∗ 2.2 − 2.2 : −2 0 ≃ −2 0 : −2 0 : −2 0 949
n = 12 3 8∗ 2.2.3 0 →
8∗ − 2 0. − 2 0.3 0 ≃ 9∗ . − 2 − 1 →
9∗ .1 ≃ 9∗ 940
n = 12 3 9∗ .(2, 2) 0 →
9∗ .2 − 2 0. − 2 − 1 ≃ 9∗ 940
n = 12 3 9∗ (2, 2) 0 →
9 2 − 2 0 : −2 − 10 : . − 2 − 1 ≃ 8∗ − 2 − 10 : 2.2.2 →
∗
8∗ 1 : −2 0. − 2 0.2 ≃ 9∗ 940
n = 12 4 6∗ (2, 2).2 0.2.2 0 →
6∗ 2 − 2. − 2.2. − 2 ≃ −2 0 : −2 0 : −2 0 949
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
n = 12 1 12n257 2 : (−2, −2 − 1) 0 : −2 0 →
2 : (−2, 1) 0 : −2 0 ≃ −2 0 : −2 0 : −2 0 949
n = 12 1 12n414 −2 − 1 0.3.2.2 0 →
1.3.2.2 0 ≃ 3 0 : 2 : 2 →
3 0 : −2 0 : −2 0 ≃ −2 0 : −2 0 : −2 0 949
n = 12 1 12n611 3.2.2 0. − 2 − 1 0 →
3.2.2 0.1 ≃ 3 0 : 2 : 2 →
3 0 : −2 0 : −2 0 ≃ −2 0 : −2 0 : −2 0 949
n = 12 1 12n745 8∗ 2.3 0. − 2 0 →
8 − 2 0.(−2, 1) 0.2 : −2 − 1 : −2 − 1 ≃ 9∗ . − 2 − 1 →
∗
9∗ .1 ≃ 9∗ 940
n = 12 1 12n760 ∗
9 : −3 0.2 0 →
9∗ : (2, −1) 0.2 0. − 2 − 1 ≃ 8∗ 2.3 0. − 2 0 →
8∗ − 2 0.(−2, 1) 0.2 : −2 − 1 : −2 − 1 ≃ 9∗ . − 2 − 1
9∗ .1 ≃ 9∗ 940
n = 12 1 12n838 −2. − 2. − 2 0.2.2.2 0 →
2 0.2 0. − 2 0. − 2 0.2. − 2 ≃ 9∗ 940
n = 12 1 12n844 8∗ 2.2 0.2 : . − 2 0 →
8∗ 2. − 2.2 : . − 2 0. − 2 − 1 ≃ 9∗ . − 4 0 →
9∗ .2 − 2 0. − 2 − 1 ≃ 9∗ 940
n = 12 1 12n847 9∗ . − 4 0 →
9∗ .2 − 2 0 ::: . − 2 − 1 ≃ 9∗ 940
n = 12 1 12n887 8∗ 2. − 3.2 →
8∗ − 2 0. − 3.2 ≃ 8∗ 2. − 2 0.2 →
8∗ − 2 0. − 2 0.2 ≃ 9∗ 940
Theorem 1.32. Among all prime knots with n ≤ 12 crossings there are
two knots with n = 9, two with n = 10, four with n = 11, and 15 with
n = 12 crossings which can not be unknotted by (2, 2)-moves. All of them
belong to the equivalence classes of knots 9∗ (940 ) and −2 0 : −2 0 : −2 0
(949 ).
118 LinKnot
120 LinKnot
Fig. 1.76
122 LinKnot
Theorem 1.35. A knot K has a bridge number 1 iff it is the unknot (see,
e.g., Gilbert and Porter, 1994, Proposition 1.4).
Proof. We accept the convention that the unknot has bridge number 1.
Suppose that k is a knot with the bridge number b(K) = 1 and let D(K)
be a diagram of K with one bridge and the smallest number of crossings
m. Hence, m ≥ 3. No arc of D(K), except the bridge, is an overpass
at any crossing. Therefore, there exist two adjacent crossings that are
endpoints of an underpass. Here we can make a Reidemeister move Ω2 and
reduce number of crossings by 2. This produces either a diagram of unknot,
or a 1-bridge diagram with a number of crossings smaller then m, which
contradicts the choice of m.
Theorem 1.36. For composite knots or links L1 and L2 the following
equality holds: b(L1 #L2 ) = b(L1 ) + b(L2 ) − 1 (Schubert, 1954).
There are two LinKnot functions dealing with direct product of KLs:
fDToDDirect which computes Dowker code of a direct product of KLs,
and fGenSignDirProd which computes signs of crossings of a direct prod-
uct of KLs in the order corresponding to Dowker code or P -data.
Theorem 1.37. For an oriented link L, there is an orientable, connected
surface S in ℜ3 with boundary L (i.e, S spans L) (Seifert, 1934; Gilbert
and Porter, 1994, Theorem 4.8; Murasugi, 1996, Theorem 5.1.1; Lickorish,
1997, Theorem 2.2; Manturov, 2004, Theorem 2.3).
This theorem was first published by F. Frankl and L. Pontryagin in
1930, and then proved by H. Seifert in 1934. His proof gives an algorithm
for constructing this surface, called Seifert surface.
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
Definition 1.65. The genus of a link L is the minimal genus of all ori-
entable surfaces which L spans (see, e.g., Gilbert and Porter, 1994; Mura-
sugi, 1996; Lickorish, 1997; Sossinsky, 2002; Manturov, 2004).
Theorem 1.39. For genus of two links L1 and L2 and their direct product
L1 #L2 , the following equality holds: g(L1 #L2 ) = g(L1 ) + g(L2 ) (Schubert,
1949).
For the proof see, e.g., the books by Murasugi (1996), Lickorish (1997),
or Manturov (2004).
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
124 LinKnot
Since the unknot has genus zero and all the other knots have genus
greater then zero, Theorem 1.34 (“Knots can not cancel each other”) is a
corollary of the last theorem.
126 LinKnot
The first non-invertible knot from Rolfsen’s knot tables is the knot 817
(.2.2 in the Conway notation). Composing two copies of this knot, two
with matching orientations, and two with different orientations, we get two
distinct composite knots which are not equivalent (Fig. 1.80). A. Kawauchi
(1979) proved that there is no deformation of the knot .2.2 that reverses
its orientation. Thirty six non-invertible knots with up to 10 crossings are
identified by R. Hartley (1983). So far, there is no general technique for
recognizing non-invertible knots.
Fig. 1.80 Different composite knots obtained from the knot 817 (.2.2).
128 LinKnot
chiral non-invertible knots. Single knot 1082 is the member of family 1).
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
130 LinKnot
Among 637 non-invertible chiral knots, 68 are obtained from the existing
families derived from knots with n = 8 and n = 10 crossings, and 554
are generators of the new families of chiral non-invertible knots without
additional conditions for parameters. In the following table, the remaining
15 chiral non-invertible knots, families of chiral non-invertible knots derived
from them, and non-invertibility conditions are given:
132 LinKnot
[0] [0] = [∞] [0] [1] = [∞] [0] [∞] = [∞, ∞] =[∞2 ]
[1] [0] = [1] [1] [1] = [0] [1] [∞] = [∞]
[∞] [0] = [0] [∞] [1] = [1] [∞] [∞] = [∞]
The right multiplications by [0] form a dihedral symmetry group with the
invariant point [1], and the right multiplications by [1] make the cyclic
group of order 3 (Fig. 1.84).
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
Fig. 1.84 (a) The right multiplications by [0]; (b) right multiplications by [1].
134 LinKnot
Stellar (pretzel) tangles are obtained from at least three R-tangles using
the operation of ramification. Pretzel KLs are numerator closures of pretzel
tangles. For denoting the types of pretzel knots double brackets will be also
omitted in the following sense: the type of pretzel knot obtained from R-
tangles of the type [0], [1], [∞], will be concisely denoted as [∞, 1, 0], since
[0]0 = [∞], [1]0 = [1], [∞]0 = [0]. For example, the type of the pretzel
tangle 2, 3, 2 1 will be [[0], [1], [∞]] = [∞, 1, 0].
For pretzel KLs with three R-tangles we have 10 possible sets of
types. [1,1,1], [1,1,∞], [0,0,1], [0,1,∞], [0,0,∞] are knots, [0,1,1], [1,∞,∞],
[0,0,0], [0,∞,∞] 2-component links, and [∞,∞,∞] are 3-component links
(Fig. 1.85). For pretzel KLs with four R-tangles, knots are [0,1,1,1],
[1,1,1,∞], [0,1,1,∞], [0,0,0,1], [0,0,1,∞], [0,0,0,∞], 2-component links
[1,1,1,1], [0,0,1,1], [1,1,∞,∞], [0,1,∞,∞], [0,0,0,0], [0,0,∞,∞], 3-component
links [1,∞,∞,∞], [0,∞,∞,∞], and 4-component links [∞,∞,∞,∞]. In gen-
eral, for pretzel KLs with t R-tangles there are t+2 2 sets of types. Among
them, t + 2 will give knots, 2t − 2 will give 2-component links, and the
number of i-component link types will be t − i + 1 (i = 3, 4, . . . , t).
For pretzel KLs consisting of more then three R-tangles, for the same
type there are several different orders17 of the particular symbols 0, 1, ∞.
For example, for the type [0,1,1,∞] there are two possible orders: [0,1,1,∞]
and [0,1,∞,1], etc.
17 Cyclic permutations of types, where reverse permutations are treated as equal.
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
In general, the following rules hold: [1,1]=[0], [1, ∞] = [∞], [a, 0] = [a]
for every a (a ∈ {0, 1, ∞}). The set of types [∞, . . . , ∞] where ∞ occurs
k times will be concisely denoted as [∞k ] (k = 1, 2, . . .). Moreover, the
calculation of the reduced types of pretzel KLs is commutative on the
symbols 0, 1, ∞. Using these rules, every set of types can be reduced to
[0], [1], or [∞k ].
The number of components of a pretzel KL of the reduced type [0] is
2, [1] are knots, and [∞k ] k-component links (k = 1, 2, . . .).
For example, [1,1,1,1,1,0,0,0,∞,∞] = [1, ∞, ∞] = [∞2 ], so it is a 2-
component link. The symbol [∞k ] represents a tangle of the type [∞] with
k − 1 already closed components. Hence, [∞k ] = [∞k−1 ], where the sub-
script k −1 denotes the number of already closed components. A numerator
closure of [∞k ] = [∞k−1 ] is a k-component link (Fig. 1.86).
Fig. 1.86 Pretzel tangle [∞, ∞, ∞, ∞] = [∞4 ] = [∞3 ] and its numerator closure giving
4-component link.
136 LinKnot
• for i = 3 the types giving knots are [1,1,1], [0,0,1], and all three R-
tangles must be mutually different;
• for i = 4 the types giving knots are [1,1,1,0], [0,0,0,1], and from the
symmetry condition follows that the R-tangles at the first and third
position must be different;
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
Fig. 1.87 (a) Non-invertible pretzel knot 5, 3, 3, 2 1; (b) invertible pretzel knot 5, 3, 2 1, 3
with mirror-symmetric t-diagram.
4 1, 2 1, 3 2 3, 2 1, 3 2 1 1, 2 2, 2 1 3 1 1, 3, 2 1
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
138 LinKnot
4 1, 2 1, 3
2 3, 2 1, 3 2 1 1, 2 2, 2 1 3 1 1, 3, 2 1
The first subset contains the knot 4 1, 2 1, 3 that belongs to the fam-
ily of non-invertible knots (2p) 1, (2q) 1, (2r + 1) with the additional non-
invertibility condition p 6= q, and the knots from the other subset generate
families of non-invertible knots without additional requirements on para-
meters.
For n = 12 there are 17 non-invertible pretzel knots:
4 2, 2 1, 3 2 4, 2 1, 3 4 1, 2 2, 3 2 2, 2 1, 5
2 1 1, 5, 3 2 1 1, 4 1, 2 1
3 1 2, 2 1, 3 2 1 2 1, 2 1, 3 2 1 1 1 1, 2 1, 3 2 3, 22, 3
3 1 1, 2 2, 3 3 2, 2 2, 2 1 2 2 1, 2 2, 2 1 2 1 1, 2 3, 21
3 1 1, 2 1 1, 2 1 2 2 1, 2 1 1, 3 3 2, 2 1 1, 3
divided into three subsets. The first subset contains non-invertible knots
obtained from the previously derived family (2p) 1, (2q) 1, (2r + 1) (p 6= q);
the second subset contains knots belonging to the new families with ad-
ditional symmetry conditions (2p) 1 1, (2q + 1), (2r + 1) (q 6= r), and
(2p) 1 1, (2q) 1, (2r) 1 (q 6= r); every member of the third subset generates
the family of non-invertible knots with no additional conditions for para-
meters.
For n = 13 we have 51 non-invertible knot, divided into the three sub-
sets:
6 1, 2 1, 3 2 5, 2 1, 3 4 3, 2 1, 3 4 1, 2 3, 3
5 1 1, 2 1, 3 3 1 1, 4 1, 3 2 3, 2 1, 5 4 1, 2 1, 5
3 1 1, 2 1, 5 4 1 1, 2 2, 2 1 2 1 1, 4 2, 2 1 2 1 1, 2 4, 2 1
2 1 1, 4 1, 2 2
3 2, 5, 3 2 2 1, 5, 3 2 4, 2 2, 3 4 2, 2 2, 3
4 1 1, 2 1 1, 3 4 1, 3 2, 2 1 2 2 1, 4 1, 2 1
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
3 1 3, 2 1, 3 3 3 1, 2 1, 3 2 1 2 2, 2 1, 3 2 2 2 1, 2 1, 3
3 2 1 1, 2 1, 3 2 1 1 1 2, 2 1, 3 2 2 1 1 1, 2 1, 3 3 1 2, 2 2, 3
2 1 2 1, 2 2, 3 2 1 1 1 1, 2 2, 3 3 1 1, 2 3, 3 2 2 1, 3 2, 3
2 1 3, 2 1 1, 3 2 3 1, 2 1 1, 3 2 1 1 2, 2 1 1, 3 3 1 1 1, 2 1 1, 3
2 1 3, 2 2, 2 1 2 3 1, 2 2, 2 1 2 1 1 2, 2 2, 2 1 3 1 1 1, 2 2, 2 1
3 2, 2 3, 2 1 2 2 1, 2 3, 2 1 3 1 1, 3 2, 2 1 3 1 2, 2 1 1, 2 1
2 1 2 1, 2 1 1, 2 1 2 1 1 1 1, 2 1 1, 2 1 3 1 1, 2 2 1, 2 1 2 1 1, 2 3, 2 2
3 1 1, 2 1 1, 2 2 2 1 1, 2 1, 3, 3 2 2, 2 1, 2 1, 3
The first subset contains knots from already derived families (with or with-
out additional symmetry conditions). Knots from the second and third
subset are the generators of new families, with and without additional sym-
metry conditions for parameters, respectively.
For n = 13 we have first non-invertible knots, 2 1 1, 2 1, 3, 3 and
2 2, 2 1, 2 1, 3, with four R-tangles, where tangles at the non-adjacent po-
sitions are different, so their t-diagrams are not mirror-symmetric. The
knots that consist of the same R-tangles, 2 1 1, 3, 2 1, 3 and 2 2, 2 1, 3, 2 1,
with the mirror-symmetric t-diagrams, are invertible.
Using LinKnot function fNinvStellar we obtained the following num-
ber of non-invertible pretzel knots with n crossings (n = 10, . . . , 17):
n 10 11 12 13 12 15 16 17
1 4 17 51 155 427 1152 2983
4 1 1, 2 1, 3+ 2 1 1, 4 1, 3+ 2 1 1, 2 1, 5+
4 1, 2 1, 3 + + 2 2, 5, 3+ 4 1, 2 2, 2 1+
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
140 LinKnot
2 1 1, 2 1, 3 + ++ 2 3, 2 1, 3 + + 3 1 1, 2 1, 3 + + 2 1 1, 2 2, 2 1 + +
2 1 3, 2 1, 3+ 2 3 1, 2 1, 3+ 2 1 1 2, 2 1, 3+ 3 1 1 1, 2 1, 3+
3 2, 2 2, 3+ 2 2 1, 2 2, 3+ 2 1 1, 2 3, 3+ 3 1 1, 2 1 1, 3+
2 3, 2 2, 2 1+ 3 1 1, 2 2, 2 1+ 2 1 1, 3 2, 2 1+ 2 2 1, 2 1 1, 2 1+
divided again into three subsets: knots belonging to already derived fam-
ilies, knots generating new families with additional symmetry conditions,
and knots generating families without additional conditions for parameters.
In order to obtain families of non-invertible pretzel knots with pluses, we
can use the same replacements as in the case of pretzel knots, but also we
can replace every sequence of k pluses with a sequence of the same parity.
The number of components of a pretzel KL with pluses can be computed
using the same rules as for pretzel KLs.
Conjecture 1.7. A knot of the form (r1 , r2 , . . . , ri )(r1′ , r2′ , . . . , rj′ ) of the
type [0] [0], [1] [∞], [∞] [1], [∞] [∞] is non-invertible iff none of its pretzel
parts is mirror-symmetric; it is chiral non-invertible iff i = j and rk = rk′
(k = 1, . . . , i), and chiral non-invertible otherwise. A knot of the type
[0] [1], [1] [0] is chiral non-invertible iff its pretzel part of the type [0] is not
mirror-symmetric.
142 LinKnot
tangles, and k = 1, 2, . . . is a knot iff its type is [0] [0] [1], [1] [0] [0], [1] [0] [∞],
[∞] [0] [1], [0] [0] [∞], or [∞] [0] [0]. From the symmetry reasons, it is suffi-
cient to consider knots of the type [0] [0] [1], [1] [0] [∞], [0] [0] [∞]. A knot of
this form is non-invertible iff it does not contain a [∞] tangle, and invertible
otherwise. Hence, all non-invertible knots are of the type [0] [0] [1].
An arborescent KL of the form (r1 , r2 ), r3 , (r4 , r5 ), where ri (i =
1, . . . , 5) are R-tangles, is a knot iff its (pretzel) type is [0, ∞, 0], [0, ∞, 1],
[1, ∞, 1], [0, 1, 0], [1, 1, 1], [∞, 1, 0], [∞, 1, 1], [1, 0, ∞], [0, 0, ∞], [0, 0, 1]. A
knot of the type [0, ∞, 0], [0, ∞, 1], [1, ∞, 1] (with ∞ in the middle) is non-
invertible iff the tangles (r1 , r2 ), (r4 , r5 ) are different. A knot of the type
[0, 1, 0], [1, 1, 1], [∞, 1, 0], [∞, 1, 1] (with 1 in the middle) is non-invertible iff
the tangle (r1 , r2 ) is different from (r4 , r5 ) and from its reverse (r5 , r4 ). A
knot of the type [1, 0, ∞], [0, 0, ∞], [0, 0, 1] (with 0 in the middle) is always
non-invertible.
The next class we consider are KLs of the form p1 , p2 , . . . , pi , where pk
(k = 3, 4, . . .) are pretzel tangles of the form (r1 , r2 , . . . , rj ) (j = 2, 3, . . .).
The result is a knot iff among pk tangles there is an odd number of the
tangles of the pretzel type [1] and an arbitrary number of the tangles of the
pretzel type [∞], or if there is exactly one tangle of the pretzel type [0]. The
simplest case of such knots are those of the form: (r1 , r2 ), (r3 , r4 ), (r5 , r6 ).
They are non-invertible iff they do not contain equal pretzel tangles,
i.e., iff their t-diagram is not mirror-symmetric. For example, knot
(5, 2), (3, 2), (2 1, 2 1) is non-invertible, and (3, 2), (3, 2), (2 1, 2 1) is invert-
ible. Non-invertible knots (3, 4), (3, 2), (2 1, 2 1), (5, 2), (3, 2), (2 1, 2 1) be-
longing to the same family are obtained by breaking symmetry of their
corresponding t-diagrams. In general, a knot of the form p1 , p2 , . . . , pi
is non-invertible iff its t-diagram is not mirror-symmetric. For example,
knots (2 1, 3), (2 1, 3), (3, 2), (3, 2 1), (2 1, 3), (2 1, 3), (2 1, 3), (3, 2 1), (2 1, 5),
(2 1, 3), (2 1, 3), (2 1, 3), (2 1, 5), (4 1, 3) are non-invertible, and (2 1, 3), (3, 2),
(2 1, 3), (3, 2 1) is invertible.
The proposed method can be used for computing symmetry groups of
non-invertible knots, where the results obtained hold for whole classes of
non-invertible knots described above. For example, the symmetry group of
every non-invertible pretzel knot is Z2 .
[0], and 6∗ r1 .r2 = 6∗ r2 .r1 , 6∗ r1 .r2 0 = 6∗ r2 .r1 0. Knot 6∗ r1 .r2 is achiral non-
invertible if r1 = r2 , and chiral non-invertible otherwise. Knot 6∗ r1 .r2 0 is
chiral non-invertible iff r1 6= r2 .
The next step is to substituting three vertices of the same basic polyhe-
dron by R-tangles. All knots of the form 6∗ r1 .r2 : r3 derived from 6∗ 2.2 : 2
are non-invertible. Knots of the form 6∗ r1 .r2 : r3 derived from 6∗ 2.2 : 2,
and knots of the form 6∗ r1 .r2 0 : r3 0 derived from 6∗ 2.2 0 : 2 0 are non-
invertible iff r1 6= r3 . Knots of the form 6∗ r1 .r2 .r3 0 derived from 6∗ 2.2.2 0
are non-invertible iff r2 6= r3 . Knots of the following forms:
are non-invertible iff they do not contain equal R-tangles. All non-invertible
knots mentioned are chiral with the trivial symmetry group.
The results for knots obtained from the basic polyhedron 6∗ with four
R-tangles are the following:
• knots of the form 6∗ r1 .r2 .r3 .r4 derived from 6∗ 2.2.2.2 are achiral non-
invertible iff r1 = r4 and r2 = r3 , and chiral non-invertible otherwise;
• knots of the form 6∗ r1 .r2 : r3 .r4 0 derived from 6∗ 2.2 : 2.2 0 are chiral
non-invertible iff r2 6= r4 , or the tangle types of (r1 ,r3 ) are not ([0],[∞])
or ([∞],[0]), and invertible otherwise;
• knots of the form 6∗ r1 .r2 .r3 0.r4 derived from 6∗ 2.2.2 0.2 are chiral non-
invertible iff r2 6= r3 , and invertible otherwise;
• knots of the form 6∗ r1 .r2 .r3 .r4 0 derived from 6∗ 2.2.2.2 0 are always
chiral non-invertible;
• knots of the form 6∗ r1 .r2 : r3 .r4 derived from 6∗ 2.2 : 2.2 are invertible
achiral iff r1 = r3 and r2 = r4 , non-invertible achiral iff r1 = r2 and r3 =
r4 , invertible iff r1 = r2 or r1 = r4 and r2 = r3 , and chiral non-invertible
otherwise;
• knots of the form 6∗ r1 .r2 0.r3 .r4 0 derived from 6∗ 2.2 0.2.2 0 are chiral
non-invertible iff r2 6= r3 , and invertible otherwise;
• knots of the form 6∗ r1 .r2 .r3 0 : r4 derived from 6∗ 2.2.2 0 : 2 are achi-
ral non-invertible iff r1 = r3 and r2 = r4 , and chiral non-invertible
otherwise;
• knots of the form 6∗ r1 .r2 .r3 : r4 0 derived from 6∗ 2.2.2 : 2 0 are chiral
non-invertible iff r1 6= r3 , and invertible otherwise;
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
144 LinKnot
• knots of the form 6∗ r1 .r2 .r3 : r4 0 derived from 6∗ 2.2.2 : 2 0 are chiral
non-invertible iff r1 6= r3 , and invertible otherwise;
• knots of the form 6∗ r1 .r2 0 : r3 0.r4 derived from 6∗ 2.2 0 : 2 0.2 are
invertible iff r1 = r3 and tangle type of (r2 , r4 ) is not ([0],[∞]) or
([∞],[0]), r2 = r4 and tangle type of (r1 , r3 ) is not ([0],[∞]) or ([∞],[0]),
and chiral non-invertible otherwise.
146 LinKnot
(2 − 1 − 1 2, 2 1 1) (−2 1 1 2, 3 2) (−2 1 − 1 2, 3)
(−2 − 1 1 2, 5) (−2 − 1 1 − 2, 1) (−2 − 1 − 1 2, 2 2 1)
(2 1 − 1 1 1, 2 0) (2 1 − 1 − 1 1, 3 1 0) (2 − 1 1 1 1, 0)
(2 − 1 1 1 − 1, 0) (2 − 1 1 − 1 1, 2 0) (2 − 1 − 1 1 1, 2 2)
(2 − 1 − 1 1 − 1, 2 0) (2 − 1 − 1 − 1 1, 4 0) (−2 1 1 1 1, 3 1 1)
(−2 1 − 1 1 1, 2 1) (−2 − 1 1 1 1, 4 1) (−2 − 1 1 − 1 1, 2 1)
(−2 − 1 − 1 1 1, 2 3) (−2 − 1 − 1 1 − 1, 2 1) (−2 − 1 − 1 − 1 1, 2 1 2 0)
(1) the first is the KL reduction, meaning that every R-tangle (with mixed
signs) can be replaced by its reduced equivalent;
(2) the other is KL extension: each reduced tangle can be replaced by its
unreduced equivalent.
−2 1 − 2 − 2 1, 2 − 1 1 1 2 1, 2 1 − 1 4
2 − 1 2 2 − 1, 2 − 1 1 1 2 1, 2 1 − 1 4
148 LinKnot
n=4 211
n=6 3111 2112
n=8 4 112 31 13 22 1111 211 121
n = 10 5 113 41 14 32 1112 311 122
23 1121 221 131 212 11111 2111 1211
n = 12 6 114 51 15 42 1113 411 123
33 1122 321 132 312 11112 3111 1212
24 1131 231 141 222 11121 2211 1221
213 11211 2121 1311 2112 111111 21111 12111
where bold 1’s represent the crossings that need to be changed in order to
obtain tangle that reduces to −1.
n−4
The number of t−1 -tangles with n crossings is 2 2 for even n (n ≥ 4),
and 0 otherwise. LinKnot function fTanUn1 gives the list of t−1 -tangles
with n crossings (n ≥ 4).
The number of R-tangles l (l ≥ 3) in a pretzel KL is called length of
a pretzel KL. Every alternating pretzel (Montesinos) KL of the length 3
remains the same after any permutation of R tangles, i.e., r1 , r2 , r3 + k =
r1 , r3 , r2 +k = ... = r3 , r2 , r1 +k. The same holds for non-alternating pretzel
KLs of the length 3.
K. Motegi (1996) gave a proof of this theorem for knots, and I. Torisu
(1998) for links.
We propose the following conjecture:
Conjecture 1.10. For every pretzel (Montesinos) knot or link L of length
l the following inequality holds: u(L) ≥ l − 2.
For pretzel KLs given by their standard diagrams, denoted by
M (k; (p1 , q1 ), ..., (pl , ql )) (pi ≥ 0, GCD(pi , qi ) = 1, i = 1, ..., l), where
(pi , qi ) denotes R-tangle ri of slope pqii , and k is the number of pluses (see,
e.g., Burde and and Zieschang, 1985), I. Torisu (1996) proved the following
theorem:
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
150 LinKnot
• K = M (0; (p, −r), (q, s), (2mn ± 1, 2n2 )), where p, q, r, s, m, n, are
some non zero integers, m and n are coprime, and ps − rq = 1;
• L = M (0; (p, −r), (p, q), (2mn ± 1, 2n2 )), where p, q, m, n, are some
non zero integers, and m and n are coprime.
• r1 − 1 r2 is a rational unlink;
• r3 k is a t−1 -tangle,
Theorem 1.46. All alternating KLs of the form (r1 , r2 + k) r3 (r4 , r5 + l),
with the following properties:
• r4 − 1 r5 is a rational unlink;
• the tangles r2 k and r1 1 r3 l are t−1 -tangles,
and all KLs obtained from them by mutation have unlinking number 1.
For example,
(4 1 1 1 2, 2 2 2 2 2 11 1 2 2 2 + 1) 1 1 1 1 1 (2 1 3 2, 4 1 1 + 3)
is an arborescent alternating knot with unknotting number 1, because
2 1 3 2 1 1 1 4 is a rational knot with unknotting number 1 (2 1 3 2 −
1 1 1 4 = 1), and 2 2 2 2 2 1 1 1 2 2 2 1 (2 2 2 2 2 − 1 1 1 2 2 2 1 = −1), and
4 1 1 1 2 1 1 1 1 1 1 3 (4 1 1 1 2 − 1 1 1 1 1 1 3 = −1) are t−1 -tangles. In the
same way, 2-component link
(3 1 1 2 1, 4 2 1 1 12 2 + 3) 3 1 1 (5 2, 5 1 1 + 2)
has the unlinking number 1, where the crossing that need to be changed in
order to unlink it is denoted by bold number 1.
Conjecture 1.11. All algebraic alternating KLs with unlinking number
1 are:
152 LinKnot
(1) r1 , r2 , 2 k− (k ≥ 1), or
(2) r1 , r2 , r3 k 1 l− (k ≥ 1, l ≥ 1), or
(3) r1 , r2 , r3 (k + 1) l−
Proof.
(1) After a crossing change in tangle 2, KL of the form r1 , r2 , 2 k− gives:
r1 , r2 , (1, −1) k− = r1 , r2 , 0 k− = r1 , r2 , −1 = r1 − 1 r2 = 1;
(2) after a crossing change in t−1 -tangle r3 k which results in −1, KL
of the form r1 , r2 , r3 k 1 l− reduces to r1 , r2 , −11 l− = r1 , r2 , 0 l− =
r1 , r2 , −1 = r1 − 1 r2 = 1;
(3) after a crossing change in t−1 -tangle r3 k which results in −1, KL of
the form r1 , r2 , r3 (k + 1) l− reduces to r1 , r2 , −1 1 l− = r1 , r2 , 0 l− =
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
r1 , r2 , −1 = r1 − 1 r2 = 1.
For example,
Proofs of the following three theorems are omitted, since they are similar
to the last one.
• r2 k (k ≥ 1) is a t−1 -tangle,
• r3 − 1 r4 is unlink,
154 LinKnot
and all KLs obtained from them by mutation, have unlinking number 1.
(2 1 1, 2+) (2 1 1 1 1, 2−)
(5 3 11 4, 2 + 4) 2 2 (2 2 1, 4−)
(2 1 1, 2+) − (3 1 1 2 5, 2)
(4 1 1, 4 3 5 1 + 2) 6 3 4 9 1 − (2 2 1 1, 3 2)
156 LinKnot
• V −1 -unlink;
• a link derived from some V −1 -unlink by a t−1 -tangle substitution.
n = 11 8∗ 3 : .2 0
n = 12 8∗ 2 1 : .2 1 0 8∗ 2 : .2 2 0 8∗ 3 1 : .2 0 8∗ 4 : .2 0
n = 13 8∗ 2 : .3 2 0 8∗ 2 1 : .2 2 0 8∗ 2 2 1 : .2 0 8∗ 5 : .2 0
1.13 Braids
K.F. Gauss was the first to notice that braids can be used to describe
knotting phenomena18 . J.W.H. Alexander in 1923 discovered a remarkable
connection between KLs and closed braids.
18 A drawing of a braid from Gauss notebooks, with strand permutation at each height,
158 LinKnot
Fig. 1.88 (a) Braid giving as the closure trefoil knot 3; (b) braid giving as the closure
Hopf link 2.
Definition 1.73. Two braids are isotopic iff there is a smooth deformation
with fixed points from the first one to the second one.
As in the case of KLs, the piecewise-linear category and smooth cate-
gory give the same result, but further in the sequel we will use piecewise-
linear drawings. As with KLs, we do not distinguish between isotopic
braids, thinking of them as representations of the same object.
Fig. 1.89 (a) Product of two braids; (b) braid b1 and its inverse b−1
1 .
160 LinKnot
strand unit braid e such that for any braid b with the same number of
strands eb = be = b (the existence of a neutral element), and for each n-
strand braid b there exists an inverse n-strand braid b−1 whose product
with b gives the trivial (unit) n-strand braid e. Let us notice that the braid
diagrams of a braid b and its inverse braid b−1 are mirror-symmetric in a
mirror line containing the joint end of their product b × b−1 (Fig. 1.89b).
Hence, the following theorem holds:
Theorem 1.56. All n-strand braids make a group called a braid group and
denoted by (Bn ,×).
The group (Bn , ×) is not commutative: the product of two braids gen-
erally depends on the order of the factors.
A braid with a single crossing is called an elementary braid. Pictures
representing braids can be algebraically encoded. Moving along a braid
from top to bottom in successive levels, we see that the braid can be rep-
resented as the successive product of elementary braids. If in an n-strand
braid we denote the crossings of the strands si and si+1 , when si over-
crosses si+1 by bi , and by b−1
i when si+1 overcrosses si (i = 1, . . . , n − 1),
we obtain algebraic codes for braids– braid words. Expressed in terms of
braid words, the equivalence relation– isotopy of braids is described by the
following relations:
Theorem 1.57. (Markov’s Theorem) The five operations described are suf-
ficient to obtain from one closed braid representation of an oriented KL any
other closed braid representation of the same oriented KL (Birman, 1976).
162 LinKnot
Fig. 1.90 (a) Commutativity of distant braids; (b) Artin’s relation; (c) trivial identity;
(d) conjugation; (e) stabilization; (f) a sequence of Markov moves to get from one closed
braid representation of figure-eight knot the other.
Definition 1.76. Within the set of braids from which L is formed there
exist braids that have the least number of strands. Any such braid is called
the minimal braid presentation of L (or simply minimal braid), and its
number of strings is called the braid index of L.
The minimal braid is not unique since many minimal braids representing
a given link L have the same number of strands.
A braid representation corresponding to a given KL can be found us-
ing Vogel’s algorithm (1990). Probably the best explanation of Vogel’s
algorithm, using the geopolitical language from the 1990s that may remain
current till our own days, is given by A. Sossinsky (2002), so we paraphrase
his description. Let us consider a planar map determined by a projection
of a knot or link L. A country in this map is said to be in turmoil if it has
two edges that belong to two different circles labelled with arrows going
in the same direction (Fig. 1.91). In the case of the knot 3 2, only the
region T is in turmoil. An operation called perestroika can be applied to
any country in turmoil. It consists in replacing two faulty edges by two
“tongues”, one of which passes over the other, forming two new crossings.
The aim is to create a new bigonal country (not in turmoil) and several new
countries, some of which (in our example two of them) may be swallowed
up by neighbouring countries. If some non-nested Seifert circles remain,
we apply the change-of-infinity operation. Vogel’s algorithm is repeated as
long as there are regions in turmoil or some non-nested Seifert circles.
Vogel’s algorithm applied to knot 3 2 is illustrated in Fig. 1.91. It can
be used to unroll a KL and obtain its braid representation (Fig. 1.92).
The Knot 2000 (K2K) function GetBraidRep is the implementation
of Vogel’s algorithm. From an input, P -data of a KL, using Reidemeister’s
moves Ω2 it transforms the diagram until obtaining braid representation.
As an external program this function also uses the program Braid-9.0 writ-
ten by A. Bartholomew. In most cases, you can get a shorter braid word
if you first reduce input KL, given by P -data, using the function Reducti
onKnotLink, and then apply the function GetBraidRep. To get the
graphical output– braid diagram for the braid word, one can use the func-
tion ShowBraid. The inverse function KnotFromBraid produces P -data
from an arbitrary braid word.
In order to work with braids using Conway symbols of KLs as an in-
put, first you need to convert the input using the LinKnot function fCre
atePData. After that you can make experiments with KLs and look for
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
164 LinKnot
dependency relations between Conway symbols and braid words. For ex-
ample, every KL of the family p (p = 2, 3, . . .) has a braid word Ap , every
KL of the family p 1 2 (p = 2, 3, . . .) has a braid word Ap bAb, to every
KL of the family p 1 1 q corresponds a braid word Ap bAbq , etc. General
formulas of this kind can be obtained not just for rational KLs, but for all
kinds of KL families.
Braid words can be used for calculating different KL invariants. The
LinKnot function fSeifert calculates the Seifert matrix of a KL given by its
Conway symbol, Dowker code, or P -data. The function fSignat calculates
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
Definition 1.77. Among the set of braids for any KL, the minimum braid
is the one that has the following properties:
166 LinKnot
an odd letter and lower case for an even letter. A binary code for any
braid can be generated by assigning 0 for an alternating crossing and 1 for
a non-alternating crossing. With the four criteria that define a minimum
braid, there is always a unique minimum for any set of braids (Kawauchi
and Tayama, 2004, 2006; Gittings, 2004).
We will present another approach: after defining braid family repre-
sentatives (BF Rs) we will establish a correspondence between BF Rs and
families of KLs given in the Conway notation. For better understanding of
this correspondence, together with the standard Conway notation, a braid-
modified Conway notation will be introduced and used. First we define a
reduced braid word, describe a general form for all reduced braid words with
s = 2 strands, generate all families of two-strand braid words, and establish
a correspondence between them and families of KLs given in the Conway
notation. Then we consider the same problem for s ≥ 3. Some applications
of minimum braids (Gittings, 2004) and braid family representatives will
be discussed in Subsection 1.14.1. All computations are made using the
program LinKnot.
We use the standard definition of a braid and description of minimum
braids given by T. Gittings (2004). Instead of a . . . a, where a capital or
lower case letter a appears p times, we write ap ; p is the degree of a (p ∈ N ).
It is also possible to work with negative powers, satisfying the relations:
A−p = ap , a−p = Ap . A number of strands is denoted by s, and a length
of a braid word by l.
Definition 1.79. A reduced braid word is a braid word with degree of every
capital or lower case letter equal to 1.
For the minimality of reduced braids we are using the following criteria:
According to the first and second criterion minimal reduced braids are
the shortest reduced braids with the smallest possible number of different
letters among all equivalent reduced braids representing certain KL. A
binary code for braid crossings can be generated by assigning a zero for
an alternating, and one for a non-alternating crossing. Hence, priority will
be given to alternating braids, and then to braids which are closest to
alternating. Analogous minimality criteria can be applied to source braids.
Definition 1.81. Among the set of all braid families representing the same
KL family, the braid family representative (BF R) is the one that has the
following properties:
168 LinKnot
single BF R Ap bAq bAbr . Otherwise, using the minimum braid criteria (Git-
tings, 2004), the knot .3.2.2 0 will be obtained from the family Ap bAbq Abr ,
three-component link .2.3.2 0 will be obtained from Ap bAq bAbr , and the
knot .2.2.3 0 will be obtained from Ap bAq br Ab for p = 3, q = 2, r = 2.
Source braids corresponding to the families Ap bAbq Abr , Ap bAq bAbr and
Ap bAq br Ab are A2 bAb2 Ab2 , A2 bA2 bAb2 and A2 bA2 b2 Ab, respectively, and
the second source braid is minimal. Hence, the representative of the KL
family .r.p.q 0 is BF R Ap bAq bAbr .
According to this, single family of KLs given in the Conway notation
can be associated to every BF R and vice versa.
Notice that families of KLs obtained from BF Rs can overlap on a
finite number of KLs at their beginnings. For example, distinct BF Rs
AbAp bACbC and Ap bCbAbCb, giving KL families .2 1 : p and .p 1 : 2, re-
spectively, for p = 2 will have in common three-component link .2 1 : 2
(9311 ) mentioned above. According to the second BF R criterion, it will
be derived from the minimum generating braid AbAbACbC, and not from
AbACbACb. Hence, BF R AbAp bACbC giving KLs of the form .2 1 : p
starts for p = 2, and Ap bCbAbCb giving KLs of the form for .p 1 : 2 starts
for p = 3. In this way, all ambiguities can be avoided.
Every KL is algebraic if its basic polyhedron is 1∗ or polyhedral oth-
erwise. According to this criterion, all KLs are divided into two main
categories: algebraic and polyhedral. Since the correspondence between
members of a BF R and KLs is one-to-one, we can introduce the following
definition:
In this case, all KLs derived from the basic polyhedron .1 (with the
Conway symbols beginning with a dot) will be polyhedral KLs, because
their minimum crossing number representations are polyhedral.
We will consider only BF Rs corresponding to prime KLs.
Every 1-strand BF R is of the form Ap , with the corresponding KL
family p in the Conway notation.
Theorem 1.59. All algebraic alternating KLs with s = 2 are the members
of the following families:
Minimum braids include one additional braid (Ap bq Abr ) in the case of
algebraic alternating KLs with s = 2.
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
170 LinKnot
Table 1
Basic polyhedron .1 = 6∗
Ap bAbAb .p (1) Ap bAbAq br r : p0 : q0 (7)
Ap bAbAbq .p.q (2) Ap bAbq Ar bs p.s.r.q (8)
Ap bAq bAb .p.q 0 (3) Ap bAq bAr bs q 0.p.r 0.s 0 (9)
Ap bAbq Ab .p : q 0 (4) Ap bAq br Abs .p.s.r 0.q 0 (10)
Ap bAq bAbr .r.p.q 0 (5) Ap bAq br As bt p.t.s.r.q (11)
Ap bAq bAr b p:q:r (6) Ap bq Ar bs At bu p.q.r.s.t.u (12)
Basic polyhedron 8∗
Ap bAbAbAb 8∗ p Ap bAq bAbr Abs 8∗ p : q : .r : s
Ap bAbAbAbq 8∗ p.q Ap bAbq Ar bAbs 8∗ p.s : .r.q
Ap bAq bAbAb 8∗ p : q Ap bAq bAr bAs b 8∗ p : s : r : q
Ap bAbAbq Ab 8∗ p : .q Ap bAbAq br As bt 8∗ p.t.s.r.q
Ap bAbAq bAb 8∗ p :: q Ap bAq bAbr As bt 8∗ p.t.s.r : .q
Ap bAq bAbAbr 8∗ p.r :: .q Ap bAq br As bAbt 8∗ p : q.r.s : .t
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
Basic polyhedron 6∗
172 LinKnot
Basic polyhedron 8∗
For example, the first operation applied on AbAb gives AbACbC, and
the other AbAbCbC.
The (s + 1)-extending by replacement is sufficient for constructing gen-
erating minimum braids for a given s, with l = 2s, corresponding to KLs of
the form 2 . . . 2 = 2s , where 2 occurs s times. For 2 ≤ s ≤ 6 as the result we
obtain: AbAb, AbACbC, AbACbdCd, AbACbdCEdE, AbACbdCEdf Ef ,
...
The generating minimum braids for given s, with l = 3s − 2, cor-
responding to KLs of the form 2 1 . . . 1 2 = 2 13s−6 2, where 1 occurs
3s − 6 times, can be obtained using only (s + 1)-extension by addition.
For 3 ≤ s ≤ 6 we obtain: AbAbCbC, AbAbCbCdCd, AbAbCbCdCdEdE,
AbAbCbCdCdEdEf Ef , . . .
In a similar way, from A3 we obtain the series A3 BaB, A3 BaBCbC,
A BaBCbCDcD, A3 BaBCbCDcDEdE . . ., corresponding to the knots
3
3 2, 5 2, 7 2, 9 2, . . .
Starting with w1 = AbAbCbdCd and using the (s + 1)-extension by
replacement, the generating minimum braids with l = 2s+1, corresponding
to KLs of the form 2 2 1 . . . 1 2 = 22 12s−5 2 are obtained for given s.
However, for exhaustive derivation of reduced minimum braids we use
all combinations of (s + 1)-extending operations.
174 LinKnot
s = 4 l = 8 AbACbdCd 2222
s = 4 l = 9 AbAbCbdCd 221112
s = 4 l = 10 AbAbCbCdCd 21111112
176 LinKnot
178 LinKnot
For s = 4 and l ≤ 12, the polyhedral generating braids and their corre-
sponding KLs are given in the following table, with the notation for basic
polyhedra with 12 crossings according to A. Caudron (1982):
l = 10 AbAbACbdCd .2 2 1 l = 12 AbAbACbdCdCd 12J
l = 10 AbACbCbdCd .2 1.2 1 l = 12 AbAbACdCbCdC 11∗∗∗ : .2 0
l = 10 AbACbdCbdC .2 1 : 2 1 0 l = 12 AbAbCbAbdCbd 9∗ 2 2
l = 10 AbACdCbCdC .2 2 : 2 l = 12 AbAbCbCdCbCd 8∗ 2 1 1 :: 2 0
l = 12 AbAbCbdCbCdC 8∗ 2 1 1 0 : .2 0
l = 11 AbAbACbCdCd .2 1 1 1 1 l = 12 AbAbCbdCbdCd 9∗ 2 1 1
l = 11 AbAbCbCbdCd .2 1 1.2 1 0 l = 12 AbAbCdCbCdCd 8∗ 2 1 1 1 0
l = 11 AbAbCbdCbdC .2 1 1 : 2 1 l = 12 AbACbAdCbdCd 12L
l = 11 AbAbCdCbCdC .2 1 1 1 : 2 l = 12 AbACbCbCbdCd 8∗ 2 1 0.2 1 0
l = 11 AbACbACbdCd 9∗ 2 1 0 l = 12 AbACbCbdCbCd 9∗ .2 1 : .2
l = 11 AbACbCdCbCd 8∗ 2 1 0 :: 2 0 l = 12 AbACbCbdCbdC 8∗ 2 1 0 : .2 1 0
l = 11 AbACbCdCdCd .2 2 1 1 l = 12 AbACbCdCbCdC 9∗ 2 1 : 2
l = 11 AbACbdCbCdC 8∗ 2 1 : .2 0 l = 12 AbACbCdCbdCd 10∗∗ : 2 1 0
l = 11 AbACdCbCdCd 8∗ 2 2 0 l = 12 AbACbdCbCdCd 10∗∗ .2 1
l = 12 AbACbdCbdCdC 10∗∗ : 2 1
l = 12 AbAbAbACbdCd 8∗ 2 2 1 0 l = 12 AbCbAbCdCbCd 10∗∗ : 2 0 :: .2 0
l = 12 AbAbACbAbdCd 9∗ .2 2 l = 12 AbCbACbdCbCd 10∗∗ 2 0 :: .2 0
180 LinKnot
Definition 1.86. The minimum braid unlinking gap is the positive differ-
ence between the unlinking number obtained from a minimum braid uB (L)
and BJ-unlinking number uBJ (L) of a link L, i.e.,
182 LinKnot
Rational KLs are easy to distinguish and compare: very fast and simple
function RatReduce (based on two Mathematica functions: Continued
Fraction and FromContinued Fraction) reduces KLs given in the Con-
way notation. To recognize and compare KLs from other KL-worlds, we
need more refined invariants and reduction methods. For alternating KLs,
one can use minimization of Dowker codes.
In the language of the Conway notation, product p q can be expressed as
the ramification (p,1,. . .,1), where 1 occurs q times. The flyping sequence
for a product p q is then: (p,1,. . .,1), (1,p,. . .,1), . . ., (1,1,. . .,p). Using the
same flyping algorithm for Conway symbols, we can obtain all projections
of an alternating KL given by its Conway symbol. The LinKnot function
fProjections, calculates all projections of alternating KLs given in the
Conway notation, their Conway symbols, and the overall number. Further-
more, all non-isomorphic projections of an alternating KL are output of the
function fDiffProjectionsAltKL. Obtained projections can be minimized
using the function MinDowProjAltKL. This whole process is contained
in the function MinDowAltKL which computes the minimal Dowker code
for an alternating KL. Please notice that the function MinDowAltKL
can not compute unique minimal Dowker code of an alternating polyhedral
KL that can be obtained from different basic polyhedra. In this case one
can compute all particular minimal Dowker codes, and take the minimal
one.
Two alternating prime KLs are equal (up to their mirror images) iff
their minimal Dowker codes without signs are equal. This means that
minimal Dowker codes (without signs) are sufficient for comparing two ar-
bitrary alternating knots or links L1 and L2 given by their Conway symbols.
The function SameAltProjKL compares two alternating KL projections
given by their Conway symbols. The result is 1 for isomorphic, and 0 for
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
184 LinKnot
Fig. 1.93 (a) Chiral projection of the knot (2 2, 2) (2 2, 2); (b) its achiral projection; (c)
centro-antisymmetric representation of the projection (b).
Fig. 1.94 (a) Projection of a trefoil with the period 3; (b) the same projection showing
the period 2.
Fig. 1.95 (a) The most symmetric projection of a knot 2 1 1 1 1 1 2; (b) one of its less-
symmetric projections.
The next invariant we consider, defined only for links, is the splitting
number:
186 LinKnot
knotted components into a trefoil knot. The next link with this property
is .2.2.2 0, and we propose the family of links .p 1, (p ≥ 2) with the same
property. After a crossing change, circled on the drawing (Fig. 1.97), link
.p 1 splits into an unknot and knot p 2 (obtained from the unknotted com-
ponent which contains the crossing where the change was made). We can
also define a BJ-splitting gap– difference between BJ-unlinking number
and splitting number of a link L.
Instead of analyzing splitting number and splitting gap in great detail,
we rather illustrate it with several interesting experimental results obtained
using the LinKnot function SplittNo that calculates splitting number of
a minimal projection of link given by its Conway symbol, Dowker code, or
P -data.
A family of polyhedral links .(2k), (k ≥ 1) has the BJ-unlinking number
k + 1 and splitting number 1 (see Fig. 1.96 for k = 1), so BJ-splitting
gap is at least k and can be made arbitrarily large. The same results we
obtain for the family .(2k) (2l), (k, l ≥ 1). Moreover, every link of the
form .(2a1 ) (2a2 ) . . . (2an ), (a1 , . . . , an ≥ 1) has the splitting number 1, and
P[ n+1 ]
BJ-unlinking number i=12 a2i−1 .
All our examples with splitting number 1 were 2-component links, so
the question is: is there exists a 3-component link with the splitting number
1? The answer is: probably not!
Fig. 1.96 The link .2 (or 726 ) before (a) and after crossing change (b).
Fig. 1.97 (a) The link .8 1 before (a) and after crossing change (b).
188 LinKnot
In knot theory Borromean rings are the foremost examples of links hav-
ing two remarkable properties: three mutually disjoint simple closed curves
form a link, yet no two curves are linked. Hence, if any one curve is cut,
the other two are free to separate. In the case of 3-component links these
two properties are inseparable: one follows from the other. In the case of n-
component links (n ≥ 3), n-Borromean links can be defined as n-component
non-trivial links such that any two components form a trivial link. Among
them, those with at least one non-trivial sublink, for which we will keep the
name Borromean links, will be distinguished from Brunnian links in which
every sublink is trivial (Liang and Mislow, 1994c).
It seems surprising that besides the Borromean rings, represented by
the link 632 in Rolfsen notation, no other link with the properties mentioned
above can be found in link tables (Rolfsen, 1976; Adams, 1994). The reason
for this is very simple: all existing knot tables contain only links with at
most 9 crossings. In fact, an infinite number of n-Borromean or n-Brunnian
links exist, and they can be derived as infinite series.
The first infinite series of 3-component links, beginning with the Bor-
romean rings, was discovered by P.G. Tait (1876/77b). Their geometrical
equivalent is a regular octahedron for k = 1, and (3k)-gonal antiprisms for
k ≥ 2. Their corresponding alternating links are achiral 3-Borromean links
(Fig. 1.99).
If we relax the condition that every two components do intersect, an
infinite number of “fractal” Borromean links can be derived from each n-
Borromean link. The construction is simple: it is enough to surround
an even number of the appropriately chosen crossing points of any two
components by circles (Fig. 1.100). However, our consideration will be
restricted to n-Borromean links such that each two components intersect.
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
190 LinKnot
Fig. 1.101 Infinite series of Borromean rings: (a) with bigons; (b) without bigons; (c)
after introducing new chains of bigons.
Fig. 1.102 The first infinite series of Borromean links derived from a (2n + 1)-gonal
prism.
192 LinKnot
Fig. 1.103 The other infinite series of Borromean links derived from a (2n + 1)-gonal
prism.
Fig. 1.104 (a) The first and (b) second Liang-Mislow construction.
signs, or by P -data, the function fFindCon gives its Conway symbol. For
example, Tait’s series (Fig. 1.101a) is the family .(2k + 1) : (2k + 1) 0,
(k ≥ 1). The first link in the series of Borromean links without bigons from
Figure 1.100b is the basic polyhedron 1312∗, from which originates the fam-
ily of Borromean links with bigons 1312∗ .(2k + 1) 0, (k ≥ 0) (Fig. 1.101c).
Their cutting number 1 can be checked using the function CuttNo.
Definition 1.89. A torus knot or link [m, n] is a simple closed curve on the
torus which wraps around m times meridianally and n times longitudinally.
Detailed description of torus knots is given by Murasugi (1996), in the
Chapter 7. If the integers m, n are relatively prime, the result is a torus
knot; otherwise, it is a torus link.
For a given torus knot or link [m, n], the function fTorusKL calculates
its P -data, braid word, minimal number of crossings, unknotting num-
ber, number of components, bridge number, Alexander polynomial and
(Murasugi) signature. An infinite series of torus links will be obtained for
GCD(m, n) = 3, where GCD(m, n) is the greatest common divisor for m
and n. All of them are non-alternating links derived from basic polyhe-
dra which are m-gonal antiprisms (m ≥ 3). Making them alternating, we
obtain an infinite series of Borromean links.
The main difference between “real” and “mathematical” KLs is that the
first are open-ended. We can take closed mathematical KLs and turn them
into real KLs by cutting them, fix endpoints, and compute the number of
different (non-isomorphic) classes of obtained real KLs.
For a given projection of a KL given by its Conway symbol, Dowker
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
194 LinKnot
Chapter 2
These two problems are almost inseparable and closely related with the
problem of minimizing the number of crossings of KLs. In the case of ratio-
nal KLs these problems were solved in a very simple, elegant and fast way
(in the sense of algorithm complexity), using continued fractions. Outside
the rational world various problems arise. For the unknotting problem (and
the unknotting number problem) we propose a finite algorithm based on
Bernhard-Jablan Conjecture.
Trying to solve the recognition problem, we implemented an algorithm
for recognizing alternating KLs given in Conway notation. The LinKnot
function SameAltConKL for two given alternating KLs computes their
minimal Dowker codes and compares them. In case they are equal, the
result is 1, otherwise 0.
Before we begin derivation of KLs from other worlds: stellar (or pris-
matic), arborescent stellar, arborescent generalized, and polyhedral, we will
consider some invariants, mostly polynomial, which will enable us to dis-
tinguish and recognize KLs. Those invariants can be used for eliminating
duplicates that can occur in KL derivation.
R. Fox defined an elementary invariant of KLs, called Fox’s 3-coloring
(Crowel and Fox, 1965). Given an oriented KL, denote each oriented arc
195
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
196 LinKnot
Theorem 2.1. If a link L has at least one 3-colorable diagram, then each
of its diagrams is 3-colorable. The number of 3-colorings is an invariant of
link isotopy (Livingston, 1993, page 33; Manturov, 2004, Theorem 3.5).
of a KL (page 201).
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
one 3-colorable diagram, all its diagrams are 3-colorable and the number of
3-colorings is invariant under Reidemeister moves.
Three-colorability is one of the simplest methods used in knot theory
to distinguish knots. For example, a trefoil can be distinguished from the
unknot or figure-eight knot, because the first is three-colorable, and two
others are not.
198 LinKnot
Definition 2.4. The set of different numbers of colors which can be used
for coloring L is called coloring number set of L.
is called a quandle.
Proof. In order to prove this theorem it is sufficient to show that the rules
given in the definition of quandle are satisfied after replacements a ◦ b =
b−1 ab and a/b = bab−1 . We will use the equivalent definition of quandle,
i.e., (a ◦ b)/b = a as the second rule. The following identities:
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
a ◦ a = a−1 aa = a;
(a ◦ b)/b = (b−1 ab)/b = bb−1 abb−1 = a;
(a ◦ b) ◦ c = (b−1 ab) ◦ c = c−1 b−1 abc = c−1 b−1 cc−1 acc−1 bc = (c−1 ac) ◦
−1
(c bc) = (a ◦ c) ◦ (b ◦ c),
hold for any a, b, c ∈ G, where G is arbitrary group.
To relate a quandle with KLs we define the rule of coloring (Fig. 2.3a).
Fig. 2.3 (a) Rule of coloring and invariance of quandle under (b) Ω1 ; (c) Ω2 ; (d) Ω3 .
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
200 LinKnot
Definition 2.6. Two non-isotopic knots are equivalent if one can be ob-
tained from the other by changing both the orientation of the ambient space
and that of the knot.
According to this definition, “left” and “right” trefoil are equivalent and
have isomorphic quandles.
Theorem 2.5. The knot quandle is a complete knot invariant (in the sense
of Definition 2.6) (Manturov, 2004).
One of the most important examples is the Alexander quandle, i.e., the
way of obtaining Alexander polynomial (page 211) as a quandle. Let A be
a free module over Laurent polynomial ring (with respect to a variable t)
(Definition 2.10). Then A is a quandle defined by the operations
a ◦ b = ta + (1 − t)b,
1 1
a/b = a + (1 − b).
t t
A knot or link given by its Conway symbol, Dowker code, or P -data is an
input for the LinKnot function fGenerators, which computes generators
with the list of corresponding signs of crossings. The result is a list of
ordered triples containing incoming, outgoing, and passing generator (I
O P = Incoming-Outgoing-Passing, Fig. 2.1) for each crossing, divided
according to the components of the KL.
Fig. 2.4 (a) The link 2, 2, 2 with denoted generators and its colorings with (b) k =
3(2p − 1) colors (p = 1, 2, 3, . . .); (c) k = 2p (p = 1, 2, 3, . . .) colors.
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
The function fColTest has the same input as fGenerators with the ad-
ditional number k (k ≥ 3) denoting the number of colors used for KL color-
ing. The result is the list of generators, list of their labelling, and the list of
generator colors. For example, the link 2, 2, 2 (or 631 ) is k-colorable if k = 0
(mod 2) or k = 0 (mod 3), so its color set is {2, 3, 4, 6, 8, 9, 10, 12, 14, 15, . . .}
(Fig. 2.4), where the perfect colorings are obtained for k = 0 (mod 3).
2.1.1 Group of KL
Every finitely-generated group can be given by a presentation: set of gener-
ators x1 , . . . , xm and their relations r1 , . . . , rn . For example, a cyclic group
Cn is given by the set of generators {x1 } and a single relation xn1 = e,
where n is the order of the group Cn , and e is the identity. A dihedral
group is given by the set of generators {x1 , x2 } satisfying the relations
xn1 = x22 = (x1 x2 )2 = e, or by the set of generators {y1 , y2 } satisfying the
relations y12 = y22 = (y1 y2 )n = e. Two presentations are called isomorphic
if one is algebraically equivalent to the other (see, e.g., Coxeter and Moser,
1980). From the first presentation of the dihedral group, by the substi-
tution y1 = x1 x2 we obtain the second. Presentations can be reduced in
order to obtain minimal presentations, i.e., presentations with a minimum
number of generators. For example, the presentation
(x1 , x2 , x3 : x31 = x22 = x23 = x1 x2 x−1
3 = e)
can be reduced to
(x1 , x2 : x31 = x22 = (x1 x2 )2 = e)
which is a minimal presentation of the group D3 .
A Wirtinger presentation of a link group G(L) (or link complement fun-
damental group) can be defined for every knot or link diagram L. The
group G(L) has a presentation G(L) = (x1 , x2 , . . . , xm : r1 , . . . rn ), where
x1 , . . . , xm are generators, r1 , . . . , rn are relations satisfied by generators
in crossing points, and n is a number of crossings. The rule for making
relations is: for each crossing write down the corresponding generator with
the exponent 1 if the arc is entering the crossing and −1 if it is leaving it.
For all crossings do that in the same cyclic order (“left” or “right”). If xi
is an incoming generator, xo is an outgoing, and xp is a passing generator,
we have the relation xi xp x−1 −1
o xp = e (in a “right” cycle order) (Fig. 2.1).
Beginning from a generator other then xi we obtain conjugate relation.
An m-generator presentation obtained in this way can always be reduced
to a minimal presentation. One group can have several isomorphic minimal
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
202 LinKnot
x4 x2 x−1 −1 −1 −1 −1 −1
5 x2 = e, x2 x5 x3 x5 = e, x5 x3 x1 x3 = e)
G(5) ≃ (a, b : a5 = b2 ).
This means that if L1 and L2 are ambient isotopic KLs, their corre-
sponding groups G(L1 ) and G(L2 ) will be isomorphic. The inverse state-
ment does not hold: two different KLs can have isomorphic KL-groups
(see, e.g., Crowell and Fox, 1965; Rolfsen 1976). For example, a granny
knot 3#3 and a square knot 3#(−3) have isomorphic groups, with the
same presentation (x, y, z : z −1 xz = xzx−1 , z −1 yz = yzy −1 ). For some
KL families one can explicitly describe their groups (or presentations).
For example, for every knot from the family (2k + 1) the knot group is
G((2k + 1)) = (a, b : a2 b2k+1 ); for every link from the family (2k) it is
G((2k)) = (a, b : abk a−1 b−k ),(k ≥ 1), etc.
The program SnapPea by J. Weeks can be used for computing fun-
damental groups and many other topological properties of KLs (such
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
Theorem 2.7. Each knot is either Arf equivalent to the unknot or to the
trefoil.
Arf invariant takes value 0 for unknot, and 1 for a trefoil (where “left”
and “right” trefoils are Arf equivalent).
Proof. The sketch of the proof (Kauffman, 1983; Adams, 1994, page 223;
Manturov, 2004) is the following: each Seifert surface can be thought of as a
disc with several bands attached to its boundary. Each band can be twisted
and bands can be knotted. The number of half-turn twists for each band
can be taken to be zero or one according to Arf equivalence. If we put an
orientation on the boundary of Seifert surface, then the two edges of each
band are always oppositely oriented. The Arf equivalence allows to unknot
bands: passing of one band through the other is also Arf equivalence.
Definition 2.8. The passing of one band with oriented edges through one
another is called a pass-move (Fig. 2.6a).
After unknotting bands, we can continue with the reduction of the ob-
tained surface. Since the original Seifert surface was orientable, each band
has an even number of half-twists, so we can lower the number of half
twists in each band until two or zero half-twists remain. Bands with two
half twists can be replaced by a curl. Every band must have another end
of a band between its two ends on the boundary of a disk, so if one of the
ends of each of two distinct bands l1 and l2 lie between the ends of a third
band l3 on the edge of the disk, we can slide the end of l1 along one edge
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
204 LinKnot
Fig. 2.6 (a) A pass-move; (b) four types of pieces obtained by cutting Seifert surface.
206 LinKnot
Once two out of three knot generators I, O, P are labelled, the label on
the third is forced by the consistency condition.
The following theorem holds for KL labeling:
Theorem 2.9. If a diagram of a KL can be labelled with elements from a
group G, then every diagram of the KL can be labelled with the elements
from G, regardless of the choice of orientation (Livingston, 1993).
To prove this theorem it is sufficient to check what happens to the
labelling after performing each Reidemeister move.
The use of labelling is a powerful tool for distinguishing KLs. For ex-
ample, it was very efficiently used by M. Thistlethwaite for the computer
derivation of knots. Among 12965 knots with n ≤ 13 crossings there were
only 5639 different Alexander polynomials. Using labelling from all sub-
groups of S5 , enabled him to reduce the number of unresolved cases to
about a thousand (Thistlethwaite, 1985).
Elements g and g ′ in a group G are called conjugates if there is an
element h ∈ G such that h−1 gh = g ′ . The conjugacy is an equivalence re-
lation, that preserves the cyclic structure of a permutation group G: every
element g and all its conjugates are represented as the products of permu-
tation cycles in the same way. The relation of conjugacy induces partitions
of G into equivalence classes. For example, in S5 there are seven conjugacy
classes, that can be represented by
(1), (1 2), (1 2 3), (1 2 3 4), (1 2 3 4 5), (1 2)(3 4), (1 2)(3 4 5).
If a diagram of an oriented knot can be labelled with elements of a
group G with the labels coming from a conjugacy class C of G, then every
diagram of the same knot can be labelled with the elements from C. In
the case of links, labels on each component of a labelled link belong to the
same conjugacy class.
Using consistency relationships, once a few labels are chosen, the rest
are forced. In practice, we take a knot K and fix a group G that will be
used for knot labelling. After labelling two knot generators in some crossing
by two group generators x and y, the third generator in that crossing will
be labelled by the consistency condition. Hence, each crossing determines
a label on the next arc, forced by the labels that preceded it. Equations
satisfied in G can be read from the labelling of arcs. In this way, the knot
labelling problem is reduced to solving the equations obtained from the
group G.
C. Livingston (1993) proposed an interesting example of knots 4 2 (61 )
and 3, 3, −3 (946 ), that can not be distinguished using colorings, or by
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
208 LinKnot
the most famous polynomials are Jones and Kauffman polynomial, since
they established connections between knot theory and other branches of
mathematics (algebra of operators, braid theory) and physics (statistical
models and quantum groups). Despite of all these important achievements,
there is one disappointing fact: every polynomial invariant sometimes fails,
meaning that two (or more) different KLs may have equal polynomials.
Even worse: some KLs that are really knotted are impossible to distinguish
from the unknot by some polynomial invariants. For example, there is an
infinite number of non-alternating knots with Alexander polynomial equal
to one, and an infinite number of non-alternating links with a trivial Jones
polynomial. The infinite series of non-trivial non-alternating 2-component
links:
For n = 8:
For n = 9:
• the 3-component link . − (2, 2) that has the non-minimal algebraic 10-
crossing representation (2, −2), (−2, 2), 2 is a member of the family
(2k, −2k), (−2m, 2m), 2n with trivial Alexander and Conway polyno-
mial.
For n = 10:
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
210 LinKnot
Hence, for n ≤ 10 all links with trivial Alexander and Conway polyno-
mial, except one, are members of the families with the same property. For
the link 2 0. − 2. − 2 0.2 0 we have not succeeded to find its corresponding
family of links with trivial Alexander and Conway polynomial.
Let t1 ,t2 ,. . .,tk and −t1 ,−t2 ,. . .,−tk (k ≥ 2) be rational tangles not be-
ginning nor ending with 1. All pretzel (stellar) links composed from them
have trivial Alexander and Conway polynomial. Such links are:
• 2, 2, −2, −2 for n = 8;
• −3, 3, 2, −2 and −3, 2, 3, −2 for n = 10;
• −4, 4, 2, −2, −4, 2, 4, −2, −3, 3, 3, −3, −3, 3, −3, 3, −2 2, 2 2, 2, −2,
−2 2, 2, 2 2, −2, and 2, 2, 2, −2, −2, −2 for n = 12, etc.
First knots with trivial Alexander and Conway polynomial appear for
n = 11 crossings: famous Kinoshita-Terasaka mutants (Kinoshita-Terasaka
knot . − (3, 2).2 and Conway knot . − (2, 3).2, that also can be written as
.−(2 1, 2).2 0 and .−(2, 2 1).2 0, i.e., knots 11n34 and 11n42 in the Knotscape
notation). Both of them have non-minimal 12-crossing algebraic represen-
tations: (3, −2), (2, −3), 2 and (3, −2), (−3, 2), 2. They are members of the
families of knots with n = 4k + 2l + 1 crossings that have trivial Alexander
and Conway polynomial, given by their non-minimal representations((2k +
1), −2k), (2k, −(2k + 1)), 2l and ((2k + 1), −2k), (−(2k + 1), 2k), 2l, respec-
tively.
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
Questions such as: is there any non-trivial knot with trivial Jones poly-
nomial, or is there any non-trivial link with trivial Kauffman polynomial,
are still open. The invention of Vassiliev’s invariants brought a new hope
that finite-order invariants will be able to classify KLs. Till now, that hope
is neither realized nor disproved.
212 LinKnot
• an /an+1 = an ;
• an ◦ an+1 = an ;
2) Transposition properties:
• (a/b)/(c/d) = (a/c)/(b/d);
• (a/b) ◦ (c/d) = (a ◦ c)/(b ◦ d);
• (a ◦ b) ◦ (c ◦ d) = (a ◦ c) ◦ (b ◦ d);
• (a ◦ b)/b = a;
• (a/b) ◦ b = a
214 LinKnot
Theorem 2.10. For each Conway algebra, there exists a unique function
W (L) on link diagrams that has a value an on the n-component unlink
diagrams and satisfies Conway skein relations. This function is invariant
on oriented links (Manturov, 2004, Theorem 5.3).
x ◦ y = αx + βy,
n−1
an = (β −1 (1 − α)) a1 , n ≥ 1.
(1) h
i = 1;
(2) hL ∪
i = (−a2 − a−2 )hLi;
(3) h0i = ah1i + a−1 hHi.
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
216 LinKnot
Fig. 2.12 Invariance of the Kauffman bracket polynomial with regard to the second and
third Reidemeister move.
1
(1) [
] = q 2 ;
1
(2) [L ∪
] = q 2 [L];
(3) [0] = q v[1] + [H].
− 12
(1) Z(•) = q
(2) Z(•G) = qZ(G)
(3) Z(
• ) = Z(
–•—
–• •
) + vZ(
–• )
The first rule is just an initial condition for a graph with a single vertex.
According to the second rule, adding new isolated vertex to a graph, causes
the polynomial of the graph to be multiplied by q. The third rule says
that if we pick a particular edge of a graph G, then the polynomial for G
is obtained by adding the polynomial of the graph with that edge deleted
to v times the polynomial of the graph with that edge collapsed down to a
single vertex (see, e.g., Adams, 1994).
The square bracket polynomial [L(G)] and the dichromatic polynomial
ZG (q, v) are connected by the relation:
n
ZG (q, v) = q 2 [L(G)]
218 LinKnot
Vassiliev invariants (or finite-order invariants) are the most general invari-
ants, in the sense that many KL-invariants can be deduced from them. In
a crossing change from an overcrossing to an undercrossing Vassiliev intro-
duced an intermediate phase: a catastrophe, when one part of a KL cuts
another part transversely. Besides overcrossings and undercrossings knots
or links have double points, where KL cuts itself. KLs with double points
are called singular. If we denote the set of all singular KLs by F , KLs with-
out special crossings form a subset of F denoted by Σ0 . In the same sense,
the remaining part of the set F can be divided into strata Σ1 , Σ2 , Σ3 ,. . .
consisting of singular KLs with 1,2,3,. . . double points, respectively. With
every crossing change from overcrossing to undercrossing or vice versa, a
KL becomes singular, passing through an intermediate phase– a catastro-
phe. In the same way as for ordinary KLs, we can define an ambient isotopy
for singular KLs:
Definition 2.13. Two singular knots or links L1 and L2 are ambient iso-
topic if there is an orientation-preserving homeomorphism of ℜ3 that sends
L1 to L2 preserving the arrows indicating orientation and the cyclic order
of the branches with double points.
220 LinKnot
Fig. 2.15 (a) The one-term and four-term relations; (b) the proof of four-term relation.
222 LinKnot
Fig. 2.17 The Gauss diagram of the special projection of a trefoil knot with three double
points and one loop.
The Gauss diagram of the special projection of a trefoil knot with three
double points and one loop is given in Fig. 2.17. In the same way as
before, using the one-term relation, we can delete all double points with a
loop and continue to work with proper (or reduced) special projections and
corresponding proper (or reduced) Gauss diagrams. All non-singular knots
have the same diagram– a circle without any chords. In general, many
knots correspond to the same diagram. We can construct all the different
Gauss diagrams Dn of order n (i.e., with n double points) (Fig. 2.18). Then
we can rewrite the one-term relation and four-term relation (Fig. 2.15a) in
the language of chord diagrams (Fig. 2.19a).
224 LinKnot
Fig. 2.19 (a) The one-term and four-term relations (Fig. 2.15a) in the language of
chord diagrams; (b) the actuality tables for n ≤ 3.
Using the one-term relation, we can reduce the list Dn to one diagram
for n = 2 and two diagrams for n = 3, by eliminating the diagrams d11 ,
d22 , d34 , d35 and reducing d33 to d21 . If we think of Gauss diagrams as
vectors, from the four-term relation we obtain d31 = 2d32 , so dim(D1 ) =
0, dim(D2 ) = 1, dim(D3 ) = 1, etc. Recalling that dim(D0 ) = 1, and
continuing the calculation for n = 3, 4, . . . , 9, we obtain the sequence 1,
0, 1, 1, 3, 4, 9, 14, 27, 44 of the dimensions of the spaces Dn for n =
0, 1, 2, . . . , 9 (Bar Natan, 1995). Denote the space of chord diagrams with n
chords modulo the one-term and four-term relations by An . The main result
of the combinatorial theory of Gauss diagrams and Vassiliev invariants is
expressed in Kontsevich’s theorem:
As the final result, for every n we can obtain the actuality table An : the
list of all independent Gauss diagrams with n chords (Fig. 2.19b). As the
value of n increases, the amount of computer resources necessary for the
computation of actuality table grows exponentially.
Vassiliev invariants provide the universal language to talk about quan-
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
The program LinKnot can also be used for educational purposes. Exper-
imenting with KLs and computing their polynomials, students can guess,
recognize, or rediscover some well known KL properties. For example,
to every alternating KL corresponds an Alexander polynomial with al-
ternating signs. Hence, the Alexander polynomial of any alternating KL
can not be equal to 1, but there are many non-trivial KLs with the
Alexander and Conway polynomials equal to 1. In fact, there are fami-
lies of KLs with this property (see page 209). Many different KLs have
equal Alexander polynomials. Alexander polynomial does not distinguish
a KL from its mirror image. Alexander polynomial satisfies the relation
A(K1 #K2 ) = A(K1 )A(K2 ), and so do Jones and HOMFLYPT polyno-
mials. Jones polynomials of a link L and its mirror image L satisfy the
relation VL (t) = VL (t−1 ). The HOMFLYPT polynomial of L is obtained
by substituting each l in the HOMFLYPT polynomial of L by l−1 . Jones
polynomial (or HOMFLYPT) polynomial of a KL remains the same after
changing orientations of KL components. The bracket polynomial of an
achiral KL must be palindromic, etc.
Despite all nice properties and the increasing sensitivity of the poly-
nomial invariants, they are still useless in some cases. For example, the
non-trivial link
9∗ 3.1. − 1. − 1.2. − 1. − 1.1. − 3
has a trivial Jones polynomial (see page 208), and the same is true for a
whole family of links (Eliahou, Kauffman and Thistlethwaite, 2003).
Even combined together, polynomials sometimes fail in detecting chi-
rality: for instance, Jones, HOMFLYPT, and Kauffman polynomials of the
knots 2 2, 3, −2 (942 ), (2 1, 2+) (3, −2), or 10∗∗ 2 0.2.2.2 0.2 0.2 0 and their
mirror images are equal, although these knots are chiral. Moreover, this
property holds for the families of alternating chiral KLs 10∗∗∗ p :: .p 0,
10∗∗∗ p :: .p 0 : .q 0.q, 10∗∗∗ p :: .p 0 : .q 0.q, 10∗∗∗ p :: .p 0 : .q.q 0,
10∗∗∗ p : .q 0.q.p 0, 10∗∗∗ p.q 0 : .q : .p 0.r 0.r, 10∗∗∗ p : .q 0.q.p 0 : .r.r 0, etc.
LinKnot can be used for extensive computations which may lead to
interesting discoveries. For example, after computing that the Alexander
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
226 LinKnot
• (2k + 1, −2k), (−(2k + 1), 2k), . . . , (2k + 1, −2k), (−(2k + 1), 2k), 2l
where (2k + 1, −2k), (−(2k + 1), 2k) repeats an arbitrary number of times.
All mutant knots derived from such families also have trivial Alexander and
Conway polynomials. So, maybe there still are some undiscovered proper-
ties of Alexander and Conway polynomial? The important open question
is: why are Alexander and Conway polynomial unable to distinguish knots
belonging to the same family, or even unable to distinguish some families
of knots from the unknot?
In an arbitrary family of KLs, their corresponding polynomials are well
ordered. For example, for the family of knots (2k + 1), (k ≥ 1) we obtain
the sequence of Alexander polynomials 1-t+t2 , 1-t+t2 -t3 +t4 , etc., so for
the knot family (2k + 1) we have the general formula for the Alexander
polynomial
X2k
∆(p) = (−1)i ti .
i=0
Following the same idea, for the knot families that originated from the
general Conway symbol p q, for (m, n ≥ 1, m ≥ n) we obtain
∆((2m) (2n)) = mn − (2mn + 1)t + mnt2
2n−1
X
∆((2m + 1) (2n)) = (m + 1) + (2m + 1) (−1)i ti + (m + 1)t2n
i=1
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
2m−1
X
∆((2m) (2n + 1)) = (n + 1) + (2n + 1) (−1)i ti + (n + 1)t2m .
i=1
For the knot families derived from the general Conway symbol p 1 q (m, n ≥
1, m ≥ n) we have
∆((2m+1) 1 (2n+1)) = (m+1)(n+1)−(2mn+2m+2n+1)t+(m+1)(n+1)t2
2n+2
X
∆((2m) 1 (2n + 1)) = m + (2m + 1) (−1)i ti + mt2n+3
i=1
2m+2
X
∆((2m + 1) 1 (2n)) = n + (2n + 1) (−1)i ti + nt2m+3 ,
i=1
etc.
For the mentioned family of pretzel knots (2p + 1), (2q + 1), −(2r + 1)
∆((2p + 1), (2q + 1), −(2r + 1)) =
The first world we have derived is the linear world (or L-world) that consists
of KLs given by a general Conway symbol p (p ≥ 1). For an odd p we obtain
knots, and for an even p two-component links. All knots of the L-world
are periodic with graph symmetry group G = [2, p], and knot symmetry
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
228 LinKnot
The next world is the arborescent world. Its members are multiple
combinations of KLs belonging to the preceding worlds. For example, we
have the following combinations: stellar-rational subworld (S(R)-subworld)
obtained by replacing bigons in stellar source KLs by rational tangles which
do not begin with 1, rational-stellar subworld (R(S)-subworld) obtained by
replacing the first and last bigon in rational source KLs by stellar KLs,
(R(S))(R)-subworld obtained by replacing bigons belonging to stellar parts
of the source KLs in the R(S)-subworld by rational tangles which do not
begin with 1, stellar-stellar subworld (S(S)-subworld) obtained by replacing
bigons in stellar source KLs by stellar KLs, etc. In a certain sense, the
structure of KLs looks like Chinese nested spheres, where every sphere is
placed inside the preceding one.
230 LinKnot
4 is {4, 3 1 2 2, 2 1 1},
5 is {5, 4 1, 3 2, 3 1 1, 2 3, 2 2 1, 2 1 2, 2 1 1 1}, etc.
From the stellar source links 2, 2, . . . , 2, we derive all KLs of the stellar-
rational subworld (S(R)-subworld) by replacing bigons by R-tangles– ratio-
nal tangles which do not begin with 1. Stellar-rational KLs (or arborescent
stellar KLs, according to Caudron), known also as Montesinos KLs, are
derived from the source links 2, 2, . . . , 2 for 6 ≤ n ≤ 12. They are given in
Table 1.
Table 1
n = 6 2, 2, 2
n = 7 3, 2, 2
n = 8 4, 2, 2 2, 2, 2, 2
3, 3, 2
n = 9 5, 2, 2 3, 2, 2, 2
4, 3, 2
3, 3, 3
n = 10 6, 2, 2 4, 2, 2, 2
5, 3, 2 3, 3, 2, 2
4, 4, 2 3, 2, 3, 2
4, 3, 3
n = 11 7, 2, 2 5, 2, 2, 2 3, 2, 2, 2, 2
6, 3, 2 4, 3, 2, 2
5, 4, 2 4, 2, 3, 2
5, 3, 3 3, 3, 3, 2
4, 4, 3
n = 12 8, 2, 2 6, 2, 2, 2 4, 2, 2, 2, 2 2, 2, 2, 2, 2, 2
7, 3, 2 5, 3, 2, 2 3, 3, 2, 2, 2
6, 4, 2 5, 2, 3, 2 3, 2, 3, 2, 2
6, 3, 3 4, 4, 2, 2
5, 5, 2 4, 2, 4, 2
5, 4, 3 4, 3, 3, 2
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
4, 4, 4 4, 3, 2, 3
3, 3, 3, 3
A pretzel KL a1 , a2 , a3 , . . . remains the same after cyclic permuta-
tion or reversal of its sequence of tangles. Hence, from a symbol 3, 2, 2
we obtain (3, 2, 2) and (2 1, 2, 2), from 4, 2, 2 we obtain (4, 2, 2), (3 1, 2, 2),
(2 2, 2, 2), (2 1 1, 2, 2), from 3, 3, 2 we obtain (3, 3, 2), (3, 2 1, 2) = (2 1, 3, 2)
and (2 1, 2 1, 2), and then eliminate duplicates. In the same way, we obtain
all alternating stellar-rational KLs. Thus, the stellar world is completely
included in the SR-world.
Instead of repeating the derivation of S + links from the source links of
the type 2, 2, . . . , 2 + k, (k = 1, 2, . . .), they can be obtained directly from
the stellar-rational KLs, by adding an appropriate number of pluses. For
n = 7 we add one plus to stellar-rational KLs with 6 crossings, for n = 8
we derive S + links from SR-links with 6 crossings by adding two pluses,
and from SR-links with 7 crossings by adding one plus, etc. In this way, for
every n we derive KLs with k pluses from SR-links with n − k, . . ., n − 2,
n − 1 crossings (n ≥ k + 6, k = 1, 2, . . .).
The LinKnot functions fStellarBasic, fStellar, and fStellarPlus cal-
culate the number and Conway symbols of stellar and stellar-rational KLs
without and with pluses for a given number of crossings n, respectively.
If we try to calculate the number of different classes in the first column
of Table 1 (beginning with series 1, 1, 2, 3, 4, 5, 7 for n ≤ 12), the number
of source KLs, or even the number of all KLs derived from some gener-
ating KL, we encounter different combinatorial problems. For example,
the number of different classes mentioned n above is equal to the coefficient
corresponding to q n−6 in n−1
3 , where r is the Gauss polynomial
(1 − q n ) . . . (1 − q n−r+1 )
n
= .
r (1 − q r ) . . . (1 − q)
These problems belong to the theory of partitions with a given symmetry
group (P -partitions). Let P be a permutation group P on k objects and
n ≥ k be an integer. A natural number ni is assigned to every object ki
(1 ≤ ki ≤ k), where ki=1 ni = n. Two partitions defined by signed (or
P
232 LinKnot
For example, considering source KLs from the second column of Table
1, we obtain the families:
p, q, r (p ≥ q ≥ r ≥ 2) p 1, q, r (p ≥ 2, q ≥ r ≥ 2)
p q, r, s (p, q ≥ 2, r ≥ s ≥ 2) p 1, q 1, r (p ≥ q ≥ 2, r ≥ 2)
p q 1, r, s (p, q ≥ 2, r ≥ s ≥ 2) p 1 q, r, s (p ≥ q ≥ 2, r ≥ s ≥ 2)
p 1 1 1, q, r (p ≥ 2, q ≥ r ≥ 2) p q, r 1, s (p, q, r, s ≥ 2)
p 1, q 1, r 1 (p ≥ q ≥ r ≥ 2)
In the family p,q,r we have three-component links for p = q = r = 0
(mod 2), two-component links if exactly one of the numbers p, q, r is odd,
and knots if at least two of them are odd (Fig. 2.21).
Let us consider the knot family (2k + 1), 3, 3, (k ≥ 2). The next table
contains Jones polynomials for the first six knots in this family.
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
n=6
2, 2, 2−
No. of KLs: 1
n=7
3, 2, 2− 2 1, 2, 2−
No. of KLs: 2
n=8
4, 2, 2− 3, 2 1, 2− 3, 3, 2− 2 1, 2 1, 2−
2 2, 2, 2− 2 1 1, 2, 2− 3 1, 2, 2− 2, 2, 2, 2−
2, 2, 2, 2 − −
No. of KLs: 9
n=9
5, 2, 2− 4, 3, 2− 4 1, 2, 2− 4, 2 1, 2−
2 1 1, 3, 2− 3, 3, 3− 2 2, 3, 2− 3, 2 1, 2 1−
2 2, 2 1, 2− 2 1 1, 2 1, 2− 3 2, 2, 2− 23, 2, 2−
3 1 1, 2, 2− 2 2 1, 2, 2− 3 1, 3, 2− 3 1, 2 1, 2−
3, 3, 2 1− 2 1, 2 1, 2 1− 2 1 2,2,2− 2 1 1 1,2,2−
3,2,2,2− 2 1,2,2,2− 3,2,2,2−−
No. of KLs: 23
n = 10
6,2,2− 4,4,2− 5,3,2− 5,2 1,2−
4 1,3,2− 4,3,3− 4,3,2 1− 4,2 1,2 1−
4 1,2 1,2− 4 2,2,2− 2 4,2,2− 4,2 2,2−
4 1 1,2,2− 4,2 1 1,2− 5 1,2,2− 4,3 1,2−
3 2,3,2− 3 2,2 1,2− 3 1 1,3,2− 3 1 1,2 1,2−
2 3,3,2− 2 3,2 1,2− 2 2 1,3,2− 2 2 1,2 1,2−
2 2,2 2,2− 2 2,2 1 1,2− 2 1 1,2 1 1,2− 3 1,3,3−
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
234 LinKnot
For example, for n = 10, among the KLs consisting of four parts and
with two minuses there are the following pairs of equal KLs:
4, 2, 2, 2 − − = 3 1, 2, 2, 2 − −
2 2, 2, 2, 2 − − = 2 1 1, 2, 2, 2 − −
3, 3, 2, 2 − − = 2 1, 2 1, 2, 2 − −
3, 2, 3, 2 − − = 2 1, 2, 2 1, 2 − −
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
236 LinKnot
Table 2
n = 8 (2, 2) (2, 2)
n = 9 (2, 2) 1 (2, 2)
As before, the main tool in derivation of KLs from source KLs is sym-
metry. Therefore, among the source KLs from the first column of Table 2
we distinguish links with the graph symmetry group G = [2+ , 4] ((2, 2)(2, 2)
and (2, 2, 2)(2, 2, 2)), links with the symmetry group G = [2] ((2, 2, 2)(2, 2),
(2, 2, 2, 2)(2, 2), etc.), and delete duplicates. For example, from symmetry
reasons, it is clear that (2, 2) 2 1 (2, 2) = (2, 2) 1 2 (2, 2), etc.
In the same way, we continue with the derivation of S(R)-links, by
substituting bigons that belong to stellar parts of rational-stellar source
KLs by rational tangles which do not begin with 1, denoted by k.
RSR-links derived are given in the following table:
Table 3
n = 8 (2, 2) (2, 2)
From the source KLs without pluses we directly obtain the correspond-
ing KLs with pluses. For example, from the source link without pluses
(2, 2) (2, 2) we obtain source links with pluses:
(2, 2+) (2, 2) for n = 9,
(2, 2 + +) (2, 2), (2, 2+) (2, 2+) for n = 10,
(2, 2 + +) (2, 2+) for n = 11,
(2, 2 + +) (2, 2 + +) for n = 12, etc.
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
238 LinKnot
Table 4
n = 10 (2, 2), (2, 2), 2
In the same manner, replacing the first bigon in the S-tangle (2, 2) of
the source link (2, 2), (2, 2), 2 by the tangle (2, 2), we obtain the first source
link of the next subworld ((2, 2), 2), (2, 2), 2 (or ((2, 2), 2)((2, 2), 2) in a more
symmetric form) (Fig. 2.25).
From the source KLs derived we obtain remaining KLs of this world
by using rational compositions r, r, and adding pluses.
Fig. 2.25 The link ((2, 2), 2), (2, 2), 2 = ((2, 2), 2) ((2, 2), 2).
The choice of the steps in the proposed derivation and the stratifica-
tion of the worlds is made to maximally avoid overlapping and occurrence
of duplicates. Can we be sure that no duplicates remained and that the
derivation is exhaustive? The answer to the first question gives the graph-
theoretical approach used by A. Caudron (1982) (see page 65, Fig. 1.49).
To solve the problem of possible duplicates we can use graph-transformation
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
240 LinKnot
method developed by the same author, and also the LinKnot function
SameAltConKL which compares two alternating KLs and determines
their equality. For example, for the two source links ((2, 2), 2), (2, 2), 2 and
((2, 2), 2) ((2, 2), 2) shown in Fig. 2.25 the result is: yes, they are equal.
With non-alternating KLs given in Conway notation, the situation is
somewhat different. In principle, it is possible to implement exhaustive
derivation algorithms for all non-alternating algebraic KLs (similar to that
used for the derivation of non-alternating KLs of the stellar world).
From the alternating KLs belonging to the arborescent world for n = 8
we obtain the following non−alternating KLs:
n=8
(2,2) (2,2−) (2,2) −(2,2)
No. of KLs: 2
n=9
(3,2−) (2,2) (3,2) (2,2−) (2 1,2−) (2,2) (2 1,2) (2,2−)
(2,2+) (2,2−) (2,2+) −(2,2)
No. of KLs: 6
n = 10
(4,2) (2,2−) (2,2) (4,2−) (4,2) −(2,2) (3,3) (2,2−)
(3,3−) (2,2) (3,2 1) (2,2−) (2 1,3−) (2,2) (2 1,2 1) (2,2−)
(2 1,2 1−) (2,2) (3,2) (2 1,2−) (2 1,2) (3,2−) (2 2,2) (2,2−)
(2 2,2−) (2,2) (2 1 1,2) (2,2−) (2 1 1,2−) (2,2) (3,2) (3,2−)
(2 1,2) (2 1,2−) (3,3) −(2,2) (3,2 1) −(2,2) (2 1,2 1) −(2,2)
(3,2) −(2 1,2) (3 1,2) −(2,2) (2 2,2) −(2,2) (2 1 1,2) −(2,2)
(3,2) −(3,2) (2 1,2) −(2 1,2) (2,2,2) (2,2−) (2,2,2−) (2,2)
(2,2,2−) (2,2−) (2,2,2−−) (2,2) (2,2,2) −(2,2) (2,2) 2 (2,2−)
(2,2) −2 (2,2) (2,2) 2 −(2,2) (2,2),2,(2,2−) (2,2),−2,(2,2)
(2,2),2,−(2,2) (3,2+) (2,2−) (2 1,2+) (2,2−) (2,2+) (3,2−)
(2,2+) (2 1,2−) (3,2+) −(2,2) (2 1,2+) −(2,2) (2,2+) −(3,2)
(2,2+) −(2 1,2) (2,2++) (2,2−) (2,2++) −(2,2) (2,2,2) (2,2−−)
No. of KLs: 48
which do not begin or end with 1, where every sequence is identified with
its reverse. Every alternating stellar (pretzel) KL consisting from three
rational tangles is given by the following Conway symbols, where t1 , t2 , t3
are mutually different R-tangles:
t1 , t1 , t1 t1 , t1 , t2 t1 , t2 , t3
t1 , t1 , t1 , t1 t1 , t1 , t1 , t2 t1 , t1 , t2 , t2 t1 , t2 , t1 , t2 t1 , t1 , t2 , t3
t1 , t2 , t1 , t3 t1 , t2 , t3 , t4 t1 , t2 , t4 , t3 t1 , t3 , t2 , t4
From five R-tangles we obtain 28 pretzel KLs, from six 144, from seven
832, from eight 5942, etc.
From the source link (2, 2) (2, 2) we obtain 9 arborescent KLs:
(t1 , t1 ) (t1 , t1 ) (t1 , t1 ) (t1 , t2 ) (t1 , t1 ) (t2 , t2 ) (t1 , t2 ) (t1 , t2 ) (t1 , t1 ) (t2 , t3 )
(t1 , t2 ) (t1 , t3 ) (t1 , t2 ) (t3 , t4 ) (t1 , t3 ) (t2 , t4 ) (t1 , t4 ) (t2 , t3 )
Because of symmetry, the same holds for the source link (2, 2+) (2, 2+).
From the source link (2, 2+) (2, 2) we obtain 16 KLs, and the same holds
for (2, 2 + +) (2, 2):
242 LinKnot
diagram contains at most three bigons belonging to the same face. Elim-
inating bigons is achieved by inscribing a triangle to this face, with the
vertices belonging to the face edges (e.g., coinciding with their midpoints)
(Fig. 2.26). Table 5 contains KL projections with n crossings satisfying
this necessary condition, Dowker codes of the derived basic polyhedra, and
their list (Figs. 2.27-2.28).
Table 5
n=3 3 462 6∗
n=6 312 4 8 10 12 2 6 9∗
6∗ 6 8|10 12|2 4 9∗
244 LinKnot
classical notation as 221 , 31 , 421 , 51 ,...) and their direct products, by recursive
inscribing of p-gons (p ≥ 3).
246 LinKnot
(see page 57). Lists of basic polyhedra with 12 ≤ n ≤ 20 crossings were first
introduced in the program LinKnot (in the form of databases with more
than 80 000 KLs) and we hope that the proposed notation will become a
standard for denoting basic polyhedra with a large number of crossings.
The first basic polyhedron is the octahedron 6∗ or .1, with the graph
symmetry group G = [3, 4] of order 48, generated by the 4-rotation
S = (1)(2, 3, 5, 6)(4),
2-rotation
T = (1, 3)(2, 5)(4, 6)
and inversion
Z = (1, 4)(2, 5)(3, 6)
(Fig. 2.31). It is a 3-component alternating link, the famous Borromean
rings, the first non-trivial Brunnian 3-component link, with the link sym-
metry group G′ = [3+ , 4]. Introducing orientation results in the antisym-
metry group that contains a rotational antireflection, which means that
Borromean rings are achiral. This is the only basic polyhedron given by
two symbols: 6∗ or .1 (Conway, 1970).
From the basic polyhedron 6∗ (or .1) we derive source links by replacing
its vertices by bigons. First we make all different symmetry choices of n − 6
vertices (7 ≤ n ≤ 12), i.e., all different vertex bicolorings of the octahedron.
The number of vertex bicolorings can be computed using the Polya Enu-
meration Theorem (PET) (Pólya, 1937; Harary and Palmer, 1973; Pólya
and Read, 1987). For G = [3, 4],
1 6
ZG = (t + 3t41 t2 + 9t21 t22 + 6t21 t4 + 7t32 + 6t2 t4 + 8t23 + 8t6 ),
48 1
and the coefficients of
ZG (x, 1) = 1 + x + 2x2 + 2x3 + 2x4 + x5 + x6
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
248 LinKnot
Table 6
n = 7 .2 n = 11 2.2.2.2.2
2.2.2.2.2 0
n = 8 .2.2 2.2.2.2 0.2
.2.2 0 2.2.2.2 0.2 0
2.2 0.2.2.2 0
.2 : 2 2 0.2.2.2.2 0
.2 : 2 0 2.2 0.2.2 0.2
250 LinKnot
12. We can obtain the complete list of P (R)-links derived from 6∗ (or .1)
for 7 ≤ n ≤ 12 directly from Table 7, using the isomorphism mentioned
above and working with source links for n = 12 (Table 6). After that, by
replacing every k by the corresponding rational compositions we obtain all
such KLs. The symbol k ⋄ has the similar meaning as k, and it is used to
denote partitions that are not mutually equivalent according to symmetry.
For example,
.4 ⋄ .4 ⋄ denotes .2 2.2 2 and .2 2.2 1 1 (=.2 1 1.2 2),
.3 ⋄ .3 ⋄ denotes .3.3, .3.2 1 (= .2 1.3) and .2 1.2 1,
3 ⋄ : 3 ⋄ : 3 ⋄ denotes 3 : 3 : 3, 3 : 2 1 : 2 1 (= 2 1 : 3 : 2 1 = 2 1 : 2 1 : 3)
and 2 1 : 2 1 : 2 1, etc.
Table 7
n = 7 .2
n = 8 .3 .2.2
252 LinKnot
and reflection
R = (1, 3)(5, 7)(4, 8)(2)(6)
containing its axis (Fig. 2.34). We can use PET to find the number of
different symmetry choices of the vertices (i.e., vertex bicolorings of 8∗ ). In
this case,
1 8
ZG = (t + 4t21 t32 + 5t42 + 2t24 + 4t8 ),
16 1
and the coefficients of
ZG (x, 1) = 1 + x + 4x2 + 5x3 + 8x4 + 5x5 + 4x6 + x7 + x8
represent, respectively, the number of choices of n − 8 vertices for 8 ≤ n ≤
16. For 9 ≤ n ≤ 12, these choices are:
{1}; {1, 2}, {1, 3}, {1, 4}, {1, 5};
{1, 2, 3}, {1, 2, 4}, {1, 2, 5}, {1, 3, 5}, {1, 4, 7};
{1, 2, 3, 4}, {1, 2, 3, 5}, {1, 2, 3, 6}, {1, 2, 4, 5},
{1, 2, 4, 6}, {1, 2, 4, 7}, {1, 2, 5, 6}, {1, 3, 5, 7}
corresponding, respectively, to the source KLs of the form
8∗ a; 8∗ a.b, 8∗ a : b, 8∗ a : .b, 8∗ a :: b;
8∗ a.b.c, 8∗ a.b : c, 8∗ a.b : .c, 8∗ a : b : c, 8∗ a : .b : .c;
8∗ a.b.c.d, 8∗ a.b.c : d, 8∗ a.b.c : .d, 8∗ a.b : c.d,
8∗ a.b : c : d, 8∗ a : b.c : d, 8∗ a.b : .c.d, 8∗ a : b : c : d,
given in the Conway notation. The coefficients of
ZG (x, x, 1) = 1 + 2x + 12x2 + 34x3 + 87x4 + 124x5 + 136x6 + 72x7 + 30x8
give the number of different source KLs derived from 8∗ for 8 ≤ n ≤ 16. We
can divide all the obtained vertex bicolorings into equivalence classes, with
regard to their symmetry groups, and then consider only their representa-
tives. According to this, for n = 9 we have the representative 8∗ a giving 2
source links; for n = 10 the representative 8∗ a.b (8∗ a : b, 8∗ a : .b, 8∗ a :: b)
giving 3 source links; for n = 11 two representatives: 8∗ a.b.c (8∗ a : b : c,
8∗ a : .b : .c) giving 6 source links and 8∗ a.b : c (8∗ a.b : .c) giving 8 source
links; for n = 12 five representatives: 8∗ a.b.c.d (8∗ a.b : c.d, 8∗ .a : b.c : d)
giving 10 source links, 8∗ a.b.c : d (8∗ a.b : c : d) giving 16 source links,
8∗ a.b.c : .d giving 12 source links, 8∗ a : b : c : d giving 6 source links, and
8∗ a.b : .c.d giving 7 source links, where the other members of equivalence
classes are given in parentheses. The list of source links derived from these
representatives is given in Table 8:
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
254 LinKnot
Table 8
n = 9 8∗ 2 n = 11 8∗ 2.2.2 8∗ 2.2:.2
8∗ 2 0 8∗ 2.2.2 0 8∗ 2.2 : .2 0
8∗ 2.2 0.2 8∗ 2.2 0 : .2
n = 10 8∗ 2.2 8∗ 2.2 0.2 0 8∗ 2 0.2 : .2
8∗ 2.2 0 8∗ 2 0.2.2 0 8∗ 2.2 0 : .2 0
8∗ 2 0.2 0 8∗ 2 0.2 0.2 0 8∗ 2 0.2 : .2 0
8∗ 2 0.2 0 : .2
8∗ 2 0.2 0 : .2 0
the coefficients of
ZG (x, 1) = 1 + 2x + 6x2 + 12x3 + 16x4 + 16x5 + 12x6 + 6x7 + 2x8 + x9
represent, respectively, the number of different symmetry choices of n − 9
vertices for 9 ≤ n ≤ 18, and the coefficients of
ZG (x, x, 1) = 1+4x+20x2+76x3 +202x4+388x5 +509x6 +448x7 +228x8 +4x9
the number of source KLs derived from 9∗ for 9 ≤ n ≤ 18. For n ≤ 12, these
vertex choices are divided into symmetry equivalence classes and given by
their representatives. For n = 10 we have one representative 9∗ a ({1}, {2})
generating 2 source links; for n = 11 two representatives: 9∗ a.b ({1, 2},
({1, 5}) generating 4 source links, 9∗ a : b ({1, 3}, {1, 4}, {1, 9}, {2, 5})
generating 3 source links; for n = 12 three representatives: 9∗ a.b.c ({1, 2, 3},
{1, 2, 4}, {1, 2, 8}, {1, 2, 9}, {1, 4, 6}, {1, 5, 9}) generating 6 source links,
9∗ a.b : .c ({1, 2, 5}, {1, 2, 6}, {1, 3, 4}, {1, 4, 5}) generating 8 source links,
9∗ a : .b : .c ({1, 4, 7}, {2, 5, 8}) generating 4 source links. The list of the
source links derived from these representatives is given in Table 9:
Table 9
n = 10 9∗ .2 n = 12 9∗ 2.2.2 9∗ 2.2:.2 9∗ 2:.2:.2
9∗ .2 0 9∗ 2.2.2 0 9∗ 2.2 : .2 0 9∗ 2 : .2 : .2 0
9∗ 2.2 0.2 9∗ 2.2 0 : .2 9∗ 2 : .2 0 : .2 0
n = 11 9∗ 2.2 9∗ 2:2 9∗ 2.2 0.2 0 9∗ 2 0.2 : .2 9∗ 2 0 : .2 0 : .2 0
9∗ 2.2 0 9∗ 2 : 2 0 9∗ 2 0.2.2 0 9∗ 2.2 0 : .2 0
9∗ 2 0.2 9∗ 2 0 : 2 0 9∗ 2 0.2 0.2 0 9∗ 2 0.2 : .2 0
9∗ 2 0.2 0 9∗ 2 0.2 0 : .2
9∗ 2 0.2 0 : .2 0
256 LinKnot
The next member (2 × 5)∗ of the infinite class (2 × k)∗ is the basic
polyhedron 10∗ – 5-antiprism, with the graph symmetry group G = [2+ , 10]
of order 20, generated by the rotational reflection
S̃ = (1, 2, 3, 4, 5, 6, 7, 8, 9, 10)
and by reflection
R = (1)(2, 5)(3, 4)(6)(7, 10)(8, 9)
(Fig. 2.36). According to PET,
1 10
ZG = (t + 6t52 + 5t21 t42 + 4t25 + 4t10 ),
20 1
10∗ a.b ({1, 2}, {1, 3}, {1, 6}, {1, 7}, {1, 9}) generating 3 source links 10∗ 2.2,
10∗ 2.2 0, 10∗ 2 0.2 0. Taking for n = 11 the representative 10∗ 2 (10∗ 2 0) with
P ≃ {(1)}, we can obtain for n ≤ 12 all links derived from 10∗ .
The graph symmetry group G = [2, 2]+ of order 4 corresponds to the
basic polyhedron 10∗∗ (Fig. 2.37). This group is generated by two perpen-
dicular 2-rotations
S = (1, 6)(2, 7)(3, 8)(4, 9)(5, 10)
and
S1 = (1, 6)(2, 5)(3, 4)(7, 10)(8, 9).
According to PET,
1 10
ZG = (t + t21 t42 + 2t52 ),
4 1
ZG (x, 1) = 1+3x+15x2 +32x3 +60x4 +66x5 +60x6 +32x7 +15x8 +3x9 +x10 ,
258 LinKnot
links 10∗∗∗ .2 : 2 and 10∗∗∗ .2 : 2 0, 10∗∗∗ .a.b ({2, 3}, {2, 7}) generating 3
source links 10∗∗∗ .2.2, 10∗∗∗ .2.2 0, 10∗∗∗ .2 0.2 0. For n = 12, from 10∗∗∗ 2 we
derive the generating link 10∗∗∗ 3, from 10∗∗∗ .2 the generating link 10∗∗∗ .3,
and from 10∗∗∗ .2 0 the generating link 10∗∗∗ 3.
The graph symmetry group G = [1] of order 2 corresponds to the basic
polyhedron 11∗ (Fig. 2.39). This group is generated by the reflection
R = (1, 5)(2, 4)(3)(6, 10)(7, 9)(8)(11).
According to PET,
1 11
ZG = (t + t31 t42 ),
2 1
ZG (x, 1) = 1 + 7x + 31x2 + 89x3 + 174x4 + 242x5 + 242x6
260 LinKnot
and
R1 = (1, 10)(2, 8)(3, 6)(4, 7)(5)(9, 11)
corresponds to the basic polyhedron 11∗∗ (Fig. 2.40). According to PET,
1 11
ZG = (t + 2t1 t52 + t31 t42 ),
4 1
Table 10
n = 9 .(2, 2)
262 LinKnot
n = 12 .(2, 2).2.2.2 .(2, 2).2.2.2 0 .(2, 2).2.2 0.2 0 (2, 2).2.2.2 (2, 2).2.2 0.2
.2.(2, 2).2.2 0 2.(2, 2).2.2 2.(2, 2).2 0.2
.2.2.(2, 2).2 0 2.2.2 0.(2, 2)
.2.2.2.(2, 2) 0
In the same way, from the representatives 8∗ 2, 8∗ 2.2 and 8∗ 2.2 0 (Ta-
ble 8) for n = 11 we derive the generating link 8∗ (2, 2), and for n = 12
the generating links 8∗ (3 ⋆ , 2), 8∗ (2, 2)1, 8∗ (2, 2+), 8∗ (2, 2).2, 8∗ (2, 2).2 0,
8∗ 2.(2, 2) 0. From the representative 9∗ 2 (Table 9) for n = 12 we derive the
generating link 9∗ (2, 2).
For n = 12 there exist 12 basic polyhedra (Caudron, 1982), illustrated
in Fig. 2.28, given in 3D-form showing their symmetry, not always directly
visible from their graphs or Schlegel diagrams.
As we have seen, the basis of derivation for every class of KLs are source
KLs. In the case of polyhedral world, from every basic polyhedron we first
derive source KLs, and then continue the derivation by different tangle
substitutions. For the basic polyhedra that do not have rotation axes of
order 4 incident with some vertices, we used the PET (Polya Enumeration
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
Table 11
B.P. 1 2 3 4 5 6 7 8 9 10 11 12 13 14
8∗ 2 12 34 87 124 136 72 30
9∗ 4 20 76 202 388 509 448 228 74
10∗ 2 15 56 194 428 728 800 636 272 78
10∗∗ 6 53 248 874 2040 3432 3872 2956 1296 288
11∗ 14 120 688 2700 7496 14944 21312 21320 14256 5728 1088
11∗∗ 8 65 354 1370 3788 7512 10736 10700 7208 2880 576
11∗∗∗ 8 65 354 1370 3788 7512 10736 10700 7208 2880 576
121∗ 12 138 880 3990 12672 29648 50688 63480 56320 33888 12288 2080
122∗ 2 12 54 206 596 1356 2256 2836 2504 1568 576 144
123∗ 4 34 178 746 2260 5219 8792 11032 9772 2208 436
124∗ 2 18 84 368 1096 2573 4304 5451 4776 2960 1056 224
125∗ 12 138 880 3990 12672 29648 50688 63480 56320 33888 12288 2080
126∗ 14 139 900 4000 12752 29688 50848 63560 56480 33968 12352 2112
127∗ 14 139 900 4000 12752 29688 50848 63560 56480 33968 12352 2112
128∗ 12 138 880 3990 12672 29648 50688 63480 56320 33888 12288 2080
129∗ 5 22 65 139 218 252 218 139 65 22 5 1
1210∗ 6 75 440 2025 6336 14904 25344 31860 28160 17040 6144 1072
1211∗ 4 34 172 738 2224 5170 8704 10932 9680 5920 2176 430
1212∗ 6 41 196 746 2130 4697 7772 9668 8642 5465 2132 462
131∗ 10 92 612 2955 10486 27776 55392 82948 92200 73760 40448 13520 2208
132∗ 16 167 1178 5800 20752 55192 110224 165256 183600 146928 80288 26816 4224
133∗ 16 167 1178 5800 20752 55192 110224 165256 183600 146928 80288 26816 4224
134∗ 10 92 612 2955 10486 27776 55392 82948 92200 73760 40448 13520 2208
135∗ 16 167 1178 5800 20752 55192 110224 165256 183600 146928 80288 26816 4224
136∗ 14 162 1156 5750 20652 54992 109984 164856 183280 146528 80064 26656 4160
137∗ 14 162 1156 5750 20652 54992 109984 164856 183280 146528 80064 26656 4160
138∗ 14 162 1156 5750 20652 54992 109984 164856 183280 146528 80064 26656 4160
139∗ 26 312 2288 11440 41184 109824 219648 329472 366080 292864 159744 53248 8192
1310∗ 14 162 1156 5750 20652 54992 109984 164856 183280 146528 80064 26656 4160
1311∗ 16 167 1178 5800 20752 55192 110224 165256 183600 146928 80288 26816 4224
1312∗ 10 92 612 2955 10486 27776 55392 82948 92200 73760 40448 13520 2208
1313∗ 16 167 1178 5800 20752 55192 110224 165256 183600 146928 80288 26816 4224
1314∗ 26 312 2288 11440 41184 109824 219648 329472 366080 292864 159744 53248 8192
1315∗ 26 312 2288 11440 41184 109824 219648 329472 366080 292864 159744 53248 8192
1316∗ 16 167 1178 5800 20752 55192 110224 165256 183600 146928 80288 26816 4224
1317∗ 26 312 2288 11440 41184 109824 219648 329472 366080 292864 159744 53248 8192
1318∗ 14 162 1156 5750 20652 54992 109984 164856 183280 146528 80064 26656 4160
1319∗ 14 126 792 3589 12154 31240 61168 90856 100848 81392 45472 15808 2688
141∗ 10 107 768 4126 16236 48518 110464 193212 257296 257520 187264 94016 28992 4320
142∗ 16 190 1480 8062 32152 96296 219968 384824 512992 513088 373120 186784 57472 8320
143∗ 28 364 2912 16016 64064 192192 439296 768768 1025024 1025024 745472 372736 114688 16384
144∗ 14 189 1456 8050 32032 96236 219648 384664 512512 512848 372736 186592 57344 8256
145∗ 16 190 1480 8062 32152 96296 219968 384824 512992 513088 373120 186784 57472 8320
146∗ 16 190 1480 8062 32152 96296 219968 384824 512992 513088 373120 186784 57472 8320
147∗ 14 189 1456 8050 32032 96236 219648 384664 512512 512848 372736 186592 57344 8256
148∗ 14 189 1456 8050 32032 96236 219648 384664 512512 512848 372736 186592 57344 8256
149∗ 14 189 1456 8050 32032 96236 219648 384664 512512 512848 372736 186592 57344 8256
1410∗ 28 364 2912 16016 64064 192192 439296 768768 1025024 1025024 745472 372736 114688 16384
1411∗ 14 189 1456 8050 32032 96236 219648 384664 512512 512848 372736 186592 57344 8256
1412∗ 8 102 740 4073 16076 48288 109984 192692 256496 256880 186560 93616 28736 4224
1414∗ 12 128 876 4503 17244 50684 114304 198888 264304 264712 193216 97856 30784 4704
1415∗ 28 364 2912 16016 64064 192192 439296 768768 1025024 1025024 745472 372736 114688 16384
1416∗ 18 199 1512 8156 32352 96696 220608 385704 514112 514128 374144 187392 57856 8448
1417∗ 8 102 740 4073 16076 48288 109984 192692 256496 256880 186560 93616 28736 4224
1418∗ 28 364 2912 16016 64064 192192 439296 768768 1025024 1025024 745472 372736 114688 16384
1419∗ 14 189 1456 8050 32032 96236 219648 384664 512512 512848 372736 186592 57344 8256
1420∗ 14 189 1456 8050 32032 96236 219648 384664 512512 512848 372736 186592 57344 8256
1421∗ 28 364 2912 16016 64064 192192 439296 768768 1025024 1025024 745472 372736 114688 16384
1422∗ 18 199 1512 8156 32352 96696 220608 385704 514112 514128 374144 187392 57856 8448
1423∗ 18 199 1512 8156 32352 96696 220608 385704 514112 514128 374144 187392 57856 8448
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
264 LinKnot
1424∗ 8 102 740 4073 16076 48288 109984 192692 256496 256880 186560 93616 28736 4224
1425∗ 16 190 1480 8062 32152 96296 219968 384824 512992 513088 373120 186784 57472 8320
1426∗ 16 190 1480 8062 32152 96296 219968 384824 512992 513088 373120 186784 57472 8320
1427∗ 28 364 2912 16016 64064 192192 439296 768768 1025024 1025024 745472 372736 114688 16384
1428∗ 14 189 1456 8050 32032 96236 219648 384664 512512 512848 372736 186592 57344 8256
1429∗ 18 199 1512 8156 32352 96696 220608 385704 514112 514128 374144 187392 57856 8448
1430∗ 18 199 1512 8156 32352 96696 220608 385704 514112 514128 374144 187392 57856 8448
1431∗ 16 190 1480 8062 32152 96296 219968 384824 512992 513088 373120 186784 57472 8320
1432∗ 18 199 1512 8156 32352 96696 220608 385704 514112 514128 374144 187392 57856 8448
1433∗ 8 102 740 4073 16076 48288 109984 192692 256496 256880 186560 93616 28736 4224
1434∗ 28 364 2912 16016 64064 192192 439296 768768 1025024 1025024 745472 372736 114688 16384
1435∗ 28 364 2912 16016 64064 192192 439296 768768 1025024 1025024 745472 372736 114688 16384
1436∗ 28 364 2912 16016 64064 192192 439296 768768 1025024 1025024 745472 372736 114688 16384
1437∗ 18 199 1512 8156 32352 96696 220608 385704 514112 514128 374144 187392 57856 8448
1438∗ 16 190 1480 8062 32152 96296 219968 384824 512992 513088 373120 186784 57472 8320
1439∗ 28 364 2912 16016 64064 192192 439296 768768 1025024 1025024 745472 372736 114688 16384
1440∗ 28 364 2912 16016 64064 192192 439296 768768 1025024 1025024 745472 372736 114688 16384
1441∗ 16 190 1480 8062 32152 96296 219968 384824 512992 513088 373120 186784 57472 8320
1442∗ 28 364 2912 16016 64064 192192 439296 768768 1025024 1025024 745472 372736 114688 16384
1443∗ 16 190 1480 8062 32152 96296 219968 384824 512992 513088 373120 186784 57472 8320
1444∗ 14 189 1456 8050 32032 96236 219648 384664 512512 512848 372736 186592 57344 8256
1445∗ 28 364 2912 16016 64064 192192 439296 768768 1025024 1025024 745472 372736 114688 16384
1446∗ 28 364 2912 16016 64064 192192 439296 768768 1025024 1025024 745472 372736 114688 16384
1447∗ 14 189 1456 8050 32032 96236 219648 384664 512512 512848 372736 186592 57344 8256
1448∗ 6 60 398 2112 8198 24447 55472 97009 129048 129254 93984 47322 14624 2244
1449∗ 2 21 116 623 2348 7044 15850 27836 36848 37088 26816 13648 4160 687
1450∗ 14 189 1456 8050 32032 96236 219648 384664 512512 512848 372736 186592 57344 8256
1452∗ 28 364 2912 16016 64064 192192 439296 768768 1025024 1025024 745472 372736 114688 16384
1453∗ 8 102 740 4073 16076 48288 109984 192692 256496 256880 186560 93616 28736 4224
1454∗ 10 107 768 4126 16236 48518 110464 193212 257296 257520 187264 94016 28992 4320
1455∗ 14 189 1456 8050 32032 96236 219648 384664 512512 512848 372736 186592 57344 8256
1456∗ 14 189 1456 8050 32032 96236 219648 384664 512512 512848 372736 186592 57344 8256
1457∗ 8 102 740 4073 16076 48288 109984 192692 256496 256880 186560 93616 28736 4224
1458∗ 28 364 2912 16016 64064 192192 439296 768768 1025024 1025024 745472 372736 114688 16384
1459∗ 14 145 1012 5079 19132 55148 123008 212312 281440 282048 207232 106144 33920 5248
1460∗ 6 49 252 1069 3500 9314 19604 32888 42942 43299 32464 17568 6144 1176
1461∗ 24 274 1976 10050 38024 109896 245376 423744 561920 562944 413696 211456 67584 10240
1462∗ 18 175 1152 5550 20340 57714 127488 218868 289568 290280 214208 110592 36096 5760
1463∗ 10 107 768 4126 16236 48518 110464 193212 257296 257520 187264 94016 28992 4320
1464∗ 10 107 768 4126 16236 48518 110464 193212 257296 257520 187264 94016 28992 4320
8∗ 2.2 0:2 0.2 0:2 0 8∗ 2.2 0.2 0:2 0:2 0 8∗ 2.2 0:2 0:2 0.2 0 8∗ 2.2.2 0:2 0:2 0
8∗ 2.2 0.2 0:2 0.2 8∗ 2.2 0.2:2 0:2 0 8∗ 2.2 0.2 0:2 0:2 8∗ 2.2 0:2 0.2 0:2
8∗ 2.2 0:2 0.2 0.2 0 8∗ 2.2 0.2 0.2 0:2 0 8∗ 2.2 0.2 0:2.2 0 8∗ 2.2 0.2:2 0.2
8∗ 2.2 0.2:2 0.2 0 8∗ 2 0.2 0.2 0:2 0:2 0 8∗ 2 0.2 0:2 0.2 0:2 0 8∗ 2.2 0.2:2:2 0
8∗ 2.2 0:2.2 0:2 8∗ 2.2 0.2 0:2 0.2 0 8∗ 2.2 0.2 0.2:2 8∗ 2.2 0.2 0:2:2
8∗ 2.2.2 0.2 0:.2 8∗ 2.2.2.2 0:2 0 8∗ 2.2.2.2 0.2 8∗ 2.2:2.2 0.2
8∗ 2.2:2 0.2 0.2 0 8∗ 2.2.2.2:2 0 8∗ 2.2.2.2.2 0 8∗ 2.2.2 0:2 0.2
8∗ 2.2.2 0.2.2 8∗ 2.2.2 0:2.2 8∗ 2.2.2 0:.2.2 0 8∗ 2.2.2.2 0:.2
8∗ 2.2.2 0:.2.2 8∗ 2.2 0.2 0.2 0.2 8∗ 2.2.2 0.2 0.2 0 8∗ 2.2.2 0.2:.2 0
8∗ 2.2.2 0:2:.2 0 8∗ 2.2.2 0.2 0::2 0 8∗ 2 0.2 0.2 0.2 0.2 0 8∗ 2.2 0.2 0.2 0:.2
8∗ 2.2.2:2 0.2 0 8∗ 2:2:2 0.2 0.2 0 8∗ 2.2.2:2.2 0 8∗ 2:2 0.2 0.2 0.2 0
8∗ 2.2.2 0.2 0:2 8∗ 2.2:2.2 0:2 0 8∗ 2.2:2:2.2 0 8∗ 2.2.2 0.2:2
8∗ 2.2.2 0:2:2 8∗ 2.2.2:2:2 0 8∗ 2.2:2:2 0.2 0 8∗ 2.2:2.2:2 0
8∗ 2.2 0.2 0:.2:2 0 8∗ 2.2 0.2 0:.2 0.2 0 8∗ 2 0.2 0.2 0:2 0.2 0 8∗ 2 0.2 0.2 0.2 0:2 0
8∗ 2.2.2 0.2 0:.2 0 8∗ 2.2 0.2 0::2 0.2 0 8∗ 2.2 0:2 0:2 0:2 0 8∗ 2.2 0:.2 0.2 0.2 0
8∗ 2.2.2 0:2 0:.2 0 8∗ 2.2.2 0:.2 0.2 0 8∗ 2.2 0.2.2 0:2 0 8∗ 2.2 0:2 0.2.2 0
8∗ 2.2 0.2 0:2:.2 0 8∗ 2.2.2.2 0:.2 0 8∗ 2.2 0.2 0.2 0:.2 0 8∗ 2.2 0.2 0.2 0::2 0
8∗ 2.2.2 0.2.2 0 8∗ 2.2.2 0:.2 0.2 8∗ 2.2:2 0.2.2 0 8∗ 2.2 0.2.2 0:.2
8∗ 2.2 0:2:2:2 0 8∗ 2.2 0.2.2 0.2 8∗ 2.2.2.2:2 8∗ 2.2.2.2 0:2
8∗ 2.2.2 0.2:.2 8∗ 2.2:2.2 0:2 8∗ 2:2 0.2 0:2 0.2 0 8∗ 2:2 0.2 0.2 0:2 0
8∗ 2.2 0:2:2 0.2 0 8∗ 2.2 0.2 0:2:2 0 8∗ 2.2.2 0:2:2 0 8∗ 2.2 0.2:2:2
8∗ 2.2:2.2:2 8∗ 2.2.2:2:2 8∗ 2.2 0.2 0.2 0.2 0 8∗ 2.2 0:2 0:2 0.2
8∗ 2.2 0.2:2.2 0 8∗ 2.2.2 0.2::2 0 8∗ 2.2.2.2 0.2 0 8∗ 2.2.2:2.2
8∗ 2.2.2 0.2 0.2 8∗ 2.2.2:2 0.2 8∗ 2.2:2.2 0.2 0 8∗ 2.2:2 0.2 0:2 0
8∗ 2.2:2 0.2:2 0 8∗ 2.2.2 0:2 0.2 0 8∗ 2.2.2 0.2 0:2 0 8∗ 2.2 0.2 0.2 0:2
8∗ 2.2.2 0:2.2 0 8∗ 2.2.2.2 0::2 0 8∗ 2.2.2 0.2:2 0 8∗ 2.2.2 0:2 0:2
8∗ 2.2 0:2 0.2:2 8∗ 2.2:2 0.2 0.2 8∗ 2.2.2.2.2 8∗ 2.2.2:2 0:2 0
8∗ 2.2 0:2 0.2 0:.2 0 8∗ 2.2 0.2 0:2 0:.2 0 8∗ 2.2 0:2.2 0:.2 0 8∗ 2.2 0.2.2 0:.2 0
8∗ 2.2 0.2.2 0::2 0 8∗ 2.2 0.2 0:.2 0:2 0 8∗ 2.2 0:2:2 0:2 0 8∗ 2.2 0.2.2 0:2
8∗ 2.2 0.2.2 0.2 0 8∗ 2.2 0.2 0:.2 0.2 8∗ 2.2 0.2 0.2.2 0 8∗ 2.2 0:2:2 0.2
266 LinKnot
8∗ 2.2 0.2.2 0:2 0.2 0 8∗ 2.2 0.2 0:2 0.2.2 0 8∗ 2.2.2 0.2 0.2 0:.2 0 8∗ 2.2.2 0.2.2 0:2 0
8∗ 2.2 0.2.2 0.2 0.2 0 8∗ 2.2 0.2 0.2 0.2.2 0 8∗ 2.2 0.2 0:2 0.2:2 0 8∗ 2.2.2 0.2.2 0.2
8∗ 2.2.2.2 0.2.2 0 8∗ 2.2.2.2.2 0:.2 0 8∗ 2.2:2.2 0.2.2 0 8∗ 2.2 0.2.2 0:2.2 0
8∗ 2.2 0.2.2 0.2:2 0 8∗ 2.2 0.2:2 0.2.2 0 8∗ 2.2.2.2.2 0:2 8∗ 2.2.2:2.2 0.2
8∗ 2:2 0.2 0.2 0.2 0.2 0 8∗ 2.2.2.2 0.2:2 8∗ 2.2.2 0.2.2:2 8∗ 2.2.2 0.2 0:2:2 0
8∗ 2.2 0.2 0:2 0.2 0.2 8∗ 2.2 0.2 0.2 0:2 0.2 8∗ 2.2 0.2:2.2 0.2 0 8∗ 2.2 0.2 0.2:2.2 0
8∗ 2.2.2 0.2:2.2 0 8∗ 2.2:2 0.2 0.2 0.2 0 8∗ 2.2.2.2.2 0.2 0 8∗ 2.2.2:2.2.2 0
8∗ 2.2.2 0.2 0.2.2 8∗ 2.2.2 0:2.2:2 0 8∗ 2.2.2 0.2 0.2 0.2 0 8∗ 2.2.2.2 0:2 0:2 0
8∗ 2.2.2.2 0.2:.2 0 8∗ 2.2.2 0.2.2.2 0 8∗ 2.2.2.2 0:2 0.2 8∗ 2.2.2:2 0.2 0.2 0
8∗ 2.2.2 0.2 0:2.2 8∗ 2.2.2.2.2:2 0 8∗ 2.2 0.2 0.2 0.2:2 8∗ 2.2.2 0.2 0.2 0:2
8∗ 2.2.2.2:2 0.2 0 8∗ 2.2 0.2 0.2 0.2 0:.2 0 8∗ 2.2 0.2.2 0:2 0:2 0 8∗ 2.2.2 0.2.2 0:.2 0
8∗ 2.2 0.2.2 0.2.2 0 8∗ 2.2 0.2 0.2 0:.2 0.2 0 8∗ 2.2 0:2 0.2.2 0:2 0 8∗ 2.2 0.2 0.2 0:2:2 0
8∗ 2.2 0.2.2 0.2:2 8∗ 2.2.2 0.2.2 0:2 8∗ 2.2 0.2 0:2:2 0.2 0 8∗ 2.2.2.2 0.2 0:.2 0
8∗ 2.2.2 0.2 0.2.2 0 8∗ 2.2.2 0.2.2 0.2 0 8∗ 2.2 0.2.2 0:2 0.2 8∗ 2.2:2 0.2.2 0.2 0
8∗ 2.2.2:2 0.2.2 0 8∗ 2.2.2.2 0:2:2 0 8∗ 2.2.2.2.2:2 8∗ 2.2 0.2:2.2 0.2
8∗ 2.2.2:2.2.2 8∗ 2.2 0.2.2 0.2 0:.2 0 8∗ 2.2 0.2 0.2.2 0:.2 0 8∗ 2.2 0.2.2 0:2:2 0
8∗ 2.2 0.2 0.2.2 0.2 0.2 0.2 0 8∗ 2.2 0.2 0.2 0.2.2 0.2 0.2 0 8∗ 2.2.2 0.2.2 0.2 0.2.2 0
8∗ 2.2.2.2.2.2 0.2.2 0 8∗ 2.2.2.2 0.2.2 0.2 0.2 0 8∗ 2.2.2.2 0.2 0.2.2 0.2 0
8∗ 2.2.2.2.2.2.2.2 8∗ 2.2.2.2 0.2.2.2.2 0 8∗ 2.2 0.2.2 0.2.2 0.2 0.2 0
8∗ 2.2 0.2.2 0.2 0.2.2 0.2 0 8∗ 2.2.2.2 0.2.2 0.2.2 0 8∗ 2.2 0.2.2 0.2.2 0.2.2 0
As we already computed using the PET (see Table 10), from the basic
polyhedron 8∗ we derive 2 source KLs with 1 bigon, 12 with 2 bigons, 34
with 3 bigons, 87 with 4 bigons, 124 with 5 bigons, 136 with 6 bigons, 72
with 7 bigons, and 30 source KLs with 8 bigons. We can use these results
to continue with the derivation of KLs belonging to the P (R)-subworld and
to all other subworlds for larger values of n substituting bigons by R-tangles
etc).
Using the external LinKnot programs for the derivation of polyhedral
source KLs we can try to fill in the missing data in Table 10, for the
basic polyhedra 10∗∗∗ , 1413∗, and 1451∗. The main obstacles for derivation
are the size of input data: 2n source KLs for a basic polyhedron with
n crossings, and the complexity of algorithm for finding representatives
of classes of non-equivalent source KLs. So far, we were able to get the
following source KLs from the basic polyhedron 10∗∗∗ : 2 source KLs with
1 bigon, 18 with 2 bigons, 66 with 3 bigons, and 237 with 4 bigons.
The LinKnot function fGenKL generates all different alternating KLs
with n crossings from a given source KL by rational tangle substitutions.
For example, from each of P -equivalent source links 2.2.20.2, .2.2.2.20,
2.2.20.20, 2.20.2.20, 20.2.20.20 we obtained 6, 27, 100, 334, 1032, 3020, KLs
with n = 11, . . . , 16 crossings, respectively.
We are interested not only for particular polyhedral KLs with a given
number of crossings, but for their general classes, derived, e.g., from the
basic polyhedron 6∗ .
From the source link 6∗ 2 we derive 6∗ t1 ; from the source links 6∗ 2.2,
6 2.2 0, 6∗ 2 : .2 0, 6∗ 2 : .2 we obtain, respectively:
∗
6∗ t1 .t1 6∗ t1 .t2
6∗ t1 .t1 0 6∗ t1 .t2 0
6∗ t1 : .t1 0 6∗ t1 : .t2 0
6∗ t1 : .t1 6∗ t1 : .t2
From the source links 6∗ 2.2 0.2, 6∗ 2.2 : 2 0, 6∗ 2.2 0 : 2 0, 6∗ 2.2.2 0, 6∗ 2.2 :
2, 6∗ 2.2 0 :: 2 0, 6∗ 2.2.2 we obtain, respectively:
268 LinKnot
Basic polyhedra were considered first by T.S. Kirkman (who called them
“solid knots”) (1885a,b), then by J. Conway (1970), and A. Caudron (1982).
In the first part of his paper (June 2, 1984) T.S. Kirkman wrote:
“Of solid knots we are not treating. If the apparent dignity of knots
so maintains itself as to make a treatise on these n-acra desirable, it will
be no difficult thing to show in a future memoir how to enumerate and
construct them to any value of n without omission or repetition. The
beginner can amuse himself with the regular 8-hedron, which is trifilar, or
with the unifilar of eight crossings made by drawing within a square askew,
and filling up with eight triangles.”
and in its Postcript (September 1, 1984):
“As it is a brief matter, it may be worth the wile to show how all solid
knots can be constructed without omission and repetition.”
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
270 LinKnot
the first basic polyhedron with two different projections. It has two non-
isomorphic alternating diagrams– its other projection is the link 11∗∗∗ 2
(Fig. 2.48).
In this way, for example, the basic polyhedron 10∗∗∗ can be represented
as 5⋆ 5⋆ , 11∗∗∗ as 5⋆ 1 5⋆ , etc. Applying flypes to the basic polyhedra with
n ≤ 11 crossings yields nothing new– they have only one minimal alternat-
ing diagram. The first exception, basic polyhedron 12E, can be denoted
by 5⋆ , 1, 5⋆ , 1 (Fig. 2.48a). If we apply one flype, we obtain another pro-
jection 5⋆ 2 5⋆ , corresponding to the link 11∗∗∗ 2 (Fig. 2.48b). There is a
complete analogy between the first rational link 2 2 2, that has two projec-
tions, and the first basic polyhedron 12E, expressed as 5⋆ 2 5⋆ , with the
same property. In the same way, we can obtain other basic polyhedra hav-
ing more then one minimal alternating diagram, e.g., 1318∗ and 136∗ that
are two non-isomorphic projections of 5⋆ 1 1 1 5⋆ (Fig. 2.50), where 1318∗
corresponds to the projection 5⋆ 1 1 1 5⋆. Some KLs can be derived from
both, but there are KLs that can be derived only from one, but not from
some other projection of the same basic polyhedron. For example, the link
125∗ 2 0 ::: 2 0 = 5∗ , 2, 5∗ , 2 can not be obtained from the basic polyhedron
11∗∗∗ (this means, from 11∗∗∗ 2).
Among the basic polyhedra with n = 12 crossings, tree are composite:
12E = 5⋆ , 1, 5⋆ , 1 ∼ 5⋆ 2 5⋆ = 11∗∗∗ 2, 12I = 7⋆ 5⋆ , and 12J = 5⋆ 1 1 5⋆.
For n = 13 composite basic polyhedra are: 131∗ = 8⋆1 5⋆ , 135∗ = 8⋆2 5⋆ ,
1318∗ = 5⋆ 1 1 1 5⋆ ∼ 136∗ , 139∗ = 7⋆ , 1, 5⋆ , 1313∗ = 7⋆ 1 5⋆ ∼ 1311∗,
1319∗ = 5⋆ 1, 1, 5⋆, 1 ∼ (5⋆ , 1) 2 5⋆ = 1210∗2. In other words, the links
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
272 LinKnot
12E ∼ 11∗∗∗ 2, 1318∗ ∼ 136∗ , 1313∗ ∼ 1311∗, and 1319∗ ∼ 1210∗2 have
two different minimal diagrams each. All other KLs corresponding to the
basic polyhedra with n ≤ 13 crossings have a single alternating diagram.
In order to derive all composite basic polyhedra with n ≤ 15 crossings,
21 hyperbolic tangles (Fig. 2.51) we need: one with n = 5, one with n = 7,
two with n = 8, six with n = 9, and eleven with n = 10 crossings. Non-
algebraic tangles with n = 11 crossings are given in Fig. 2.52. Among basic
polyhedra with n = 14 crossings, there are 27 composite polyhedra, 18 of
their corresponding KLs permit flypes, and among them 15 will have more
then one minimal projection. Links associated to the basic polyhedra with
n = 14 crossings satisfy following equalities: 1429∗ ∼ 1430∗, 1434∗ ∼ 1445∗,
1435∗ ∼ 1439∗, 1455∗ ∼ 1456∗ ∼ 1458∗, 1463∗ ∼ 1464∗, and the links 1459∗
- 1464∗ have other projections that contain bigons. Among 76 composite
basic polyhedra with n = 15 crossings, 59 of their corresponding alternating
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
KLs permit flypes, and among 257 composite basic polyhedra with n = 16
crossings, 201 of their corresponding alternating KLs permit flypes.
Representations of composite basic polyhedra can be used for deter-
mining properties of their corresponding alternating KLs. For example,
KLs corresponding to the “palindromic” basic polyhedra 10∗∗∗ = 5⋆ 5⋆ ,
12J = 5⋆ 1 1 5⋆ , 1420∗ = 7⋆ 7⋆ , 16160∗ = 7⋆ 1 1 7⋆ are achiral from the same
reason as their analogous rational KLs from the class p p, where p is an
arbitrary tangle. The same holds for all “palindromic” basic polyhedra of
the form p′ p′ , where p′ is any non-algebraic tangle.
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
274 LinKnot
Fig. 2.48 (a) The basic polyhedron 12E; (b) the source link 11∗∗∗ 2.
Fig. 2.49 Infinite classes of non-algebraic tangles (a) starting from 5⋆ ; (b) starting from
7⋆ .
276 LinKnot
.t1 , .t1 .t2 , .t1 : t2 , .t1 .t2 .t3 , .t1 .t2 .t3 .t4 where ti (i = 1, 2, 3, 4) are algebraic
tangles (with or without zeros after the tangles). This property of the basic
polyhedra 6∗ and 10∗∗∗ is based on the fact that the non-algebraic tangle
5∗ (Fig. 2.51) has the algebraic representation ((2, −2), −2) 1, so the basic
polyhedron 6∗ can be represented in the form 5∗ , 1, i.e., as the algebraic
link ((2, −2), −2)1, 1. In the same way, 10∗∗∗ can be represented as 5∗ 5∗ ,
i.e., as the algebraic link (((2, −2), −2) 1) (((2, −2), −2) 1).
278 LinKnot
with n = 45 crossings.
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
280 LinKnot
Fig. 2.56 Basic polyhedra with n ≤ 13 crossings which have algebraic representations.
n=6 6∗ 5∗ 1
n = 10 10∗∗∗ 5∗ 5∗
n = 11 11∗∗∗ 5∗ 1 5∗
n = 12 125∗ 5∗ , 1, 5∗ , 1
1210∗ 5∗ 1 1 5∗
n = 13 136∗ (5 , 1) 1 (5∗ , 1)
∗
1318∗ 5∗ 1 1 1 5∗
1319∗ (1, 5∗ , 1) (5∗ , 1)
Among them, one of the most interesting is the basic polyhedron 1451∗,
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
Fig. 2.57 (a) A nested collapsing: basic polyhedra 1451∗ and 10∗∗∗ ; (b) knot
1451∗ : 3.2.2.2.2 0 :: .2 0 : 2 0 with n = 22 crossings and its algebraic representation
((2, −3), −(((4, −3), −3) 1) − 1) 1, ((3, −3), −3) 1 with n = 28 crossings.
which reduces into the basic polyhedron 10∗∗∗ by one 5∗ -collapse. The next
two 5∗ -collapses reduce it to a Hopf link, so this is an example of a nested
5∗ -collapsing (Fig. 2.57a). The knot 1451∗ : 3.2.2.2.2 0 :: .2 0 : 2 0 with n =
22 crossings and its algebraic representation ((2, −3), −(((4, −3), −3) 1) −
1) 1, ((3, −3), −3) 1 with n = 28 crossings is illustrated in Fig. 2.57b.
An example of a double 5∗ -collapse is the alternating knot K = 16442∗ :
.2 0 :: 2 ::: .2.2 0 with n = 20 crossings, given by the algebraic representation
with n = 26 crossings (Fig. 2.58a)
((2, (−1, −1, −(((2, −3), −3) 1))), (−1, −1, −(((2, −3), −3) 1))) 1, 1
Its basic polyhedron 16442∗ is 5∗ -collapse reducible to the knot 6∗ 2.2 = .2.2
(Fig. 2.58b), which has the algebraic representation ((2, −3), −3) 1, 1=
((2, (−1, −1, −1)), (−1, −1, −1)) 1, 1. The algebraic representation of K
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
282 LinKnot
Fig. 2.58 (a) Knot K = 16442∗ : .2 0 :: 2 ::: .2.2 0 and its algebraic representation; (b)
basic polyhedron 16442∗ which by two 5∗ -collapses reduces to .2.2.
284 LinKnot
and D) closure, and for 3-tangles there are two basic types of closures: a
closure where only adjacent vertices are joined (A-closure), and a closure
where two opposite vertices are joined (O-closure) (Fig. 2.61).
Since the result of a closure depends on a position of a 3-tangle, for every
3-tangle there are two possible A-closures and three possible O-closures.
The number of closures, i.e., the number of ways of joining 2n points on a
circle by n non-intersecting chords is known as Catalan number (or Segner
number). For n ≤ 10 Catalan numbers (the sequence A000108 from On-
Line Encyclopedia of Integer Sequences) are
n 2 3 4 5 6 7 9 9 10
Catalan no. 2 5 14 42 132 429 1430 4862 16796
2n!
In general, Catalan number is given by the formula C(n) = n!(n+1)! .
LinKnot function fAllClosures gives the list of all closures of a n-tangle
(n ≥ 3).
of elementary tangle we take the one where one strand is horizontal and
remaining n−1 strands are vertical. An elementary n-tangle |n−1| induces a
coordinate system of concentric regular 2n-gons and corresponding regions
(Fig. 2.63), where the first lower middle or right region with two vertices
is denoted by 1, and other regions (from 1 to 2n) are given in a clockwise
order. Every n-tangle placed in this coordinate system can be denoted by
|t1 t2 . . . tn−1 | (tn )r1 (tn+1 )r2 . . . (tn+k−1 )rk , where ti (i = 1, 2, . . . , n − 1)
is an algebraic tangle placed in the corresponding vertex of |n − 1| (in
the order from the right to the left), and tj rj−n+1 (rj−n+1 ∈ {1, . . . , 2n},
j = n, . . . , n + k − 1, k = 1, 2, . . .) is an algebraic tangle tj placed in the
region rj−n+1 , between k th and (k + 1)th concentric regular 2n-gon (at the
k th level). Since our primary interest is the derivation of basic polyhedra
we start adding with algebraic tangles 1 in such a way that no bigons are
created. Therefore, all pairs of adjacent regions must have different indexes.
If all algebraic tangles are 1, in order to simplify notation, instead of 1rj−n+1
we write just rj−n+1 . (Figure 2.64).
In the initial state, all (potential) algebraic tangles have the same ori-
entation (Fig. 2.65). In our notation the symbol 0 has the same meaning
as in the Conway notation for polyhedral KLs.
286 LinKnot
From every open region two arcs emerge, and adjacent regions share
the same arc. We can distinguish open regions with one, two, or more
vertices, and denote their type by 1,2,3, respectively. Placing new 1-tangle
in an open region changes its type and the types of adjacent regions. If its
original type was 1, the addition of new 1-tangle is forbidden, because a
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
288 LinKnot
• derive all basic polyhedra from k-tangles with fixed k (k ≥ 3), suffi-
ciently large to obtain all basic polyhedra with n crossings;
• for every polyhedron find minimal k such that it can be derived from
the basic k-tangle and for this k find its minimal representation.
Together with the basic polyhedra which are prime KLs, our algorithm
produces non-prime basic polyhedra that should be deleted from the lists of
basic polyhedra. The first of them is 6∗ #6∗ with the code |4| 1 3 4 3 2 1 2 4.
In the set of different codes describing the same basic polyhedron, the
minimal code is taken as the symbol of the basic polyhedron. The list
of minimal codes and their corresponding 9 basic polyhedra with n ≤ 11
crossing is given in the following table:
|2| 1 2 1 2 6∗
|2| 1 2 1 2 1 2 8∗
|2| 1 2 1 3 2 1 2 9∗
|2| 1 2 1 2 1 2 1 2 10∗
|2| 1 2 1 2 3 2 1 2 10∗∗
|3| 1 2 3 2 1 3 2 10∗∗∗
|2| 1 2 1 2 1 3 2 1 2 11∗
|2| 1 2 1 6 1 2 1 2 1 11∗∗
|3| 1 2 3 2 3 2 1 2 11∗∗∗
Among them, composite basic polyhedra 10∗∗∗ and 11∗∗∗ are derived
from the elementary 4-tangle |3|.
The list of minimal codes and their corresponding 12 basic polyhedra
with n = 12 crossings is given in the following table
290 LinKnot
6∗ (Ab)3
8∗ (Ab)4
9∗ AbACbACbC
10∗ (Ab)5
10∗∗ AbAbCbACbC
... ...
292 LinKnot
basic polyhedra of the form |n − 1| sk0 s obtained for a fixed s, and can be
denoted by (n,s,k).
Fig. 2.70 The family |2| sk0 3 2 1 2 (k ≥ 3) beginning with the basic polyhedron 9∗ .
3 6
323 343 616
3213 3216 3243 3654 6123
32123 32143 32323 32343 32654
34323 34343 34543 34654 36543
61213 61216 61616
321213 321216 321243 321323 321343
321616 323213 323243 323654 324323
324343 324543 324654 326543 343213
343243 343543 343654 345654 346543
365454 365654 612123 612323 612343
612543 612654 615654 616123 616543
3212123 3212143 3212343 3212654 3213213
3213243 3213654 3214323 3214543 3214654
3216123 3216543 3232143 3232343 3232654
3234543 3234654 3236543 3243213 3243243
3243543 3243654 3246543 3265434 3432123
3432143 3432543 3432654 3434543 3435654
3454543 3454654 3456543 3465654 3654354
3654543 6121213 6123213 6123243 6123543
6156543 6165654 6546543
32121213 32121243 32123213 32123243 32123654
32124543 32126543 32143213 32143243 32143654
32145654 32146543 32161213 32165654 32432123
61212123 61213213 61215654 61216123 61216543
61234543 61546543 61565654 61612123 61616123
61654543 65465654
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
294 LinKnot
296 LinKnot
The graph consisting of a regular 2n-gon and n chords joining its distinct
vertices is called a chord diagram of order n, or shortly n-diagram. Let the
symmetry group G act on a chord diagram.
Definition 2.25. Two n-diagrams are equivalent iff there exists an element
of the group G that transforms one to another.
The five diagrams obtained for n=3 and 17 diagrams obtained for n = 4
are illustrated in Fig. 2.72 and denoted, respectively, by 3.1-3.5 and 4.1-
4.17.
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
298 LinKnot
Fig. 2.75 Minimal sets of generators for (a) 4-diagrams; (b) 5-diagrams.
300 LinKnot
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
1 7 2 10 13 8 9 1 5 6 3 12 11 4 14 15
2 2 21 15 14 14 15 2 14 15 15 2 2 14 141 151
3 13 12 4 7 8 15 3 14 6 1 2 11 10 5 9
4 10 11 7 3 14 9 4 5 15 13 12 2 1 8 6
5 9 12 12 9 51 9 5 5 91 5 12 121 12 5 9
6 8 11 8 11 8 61 6 81 6 11 111 11 6 8 6
7 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
8 6 11 11 6 81 6 8 8 61 8 11 111 11 8 6
9 5 12 5 12 5 91 9 51 9 12 121 12 9 5 9
10 4 11 13 1 5 15 10 14 9 7 2 12 3 8 6
11 11 111 6 8 8 6 11 8 6 6 11 11 8 81 61
12 12 121 9 5 5 9 12 5 9 9 12 12 5 51 91
13 3 12 1 10 14 6 13 8 15 4 11 2 7 5 9
14 15 2 2 15 141 15 14 14 151 14 2 21 2 14 15
15 14 2 14 2 14 151 15 141 15 2 21 2 15 14 15
Fig. 2.76 3-tangle and different chord diagrams obtained from it.
Let us give a simple example how one can see that Borromean rings, i.e.,
the basic polyhedron 6∗ is a three-component link. We take its represen-
tation as the composition of two 3-tangles, compute the type of the tangle
composition, make its closure, and the conclusion is obvious: Borromean
rings are the three-component link (Fig. 2.77). Moreover, we get that
every vertex substitution in the basic polyhedron 6∗ , where all substitutes
are algebraic tangles of the type [1] gives a three-component link.
This method– tangle-type computation, enables us to determine the
number of components of any polyhedral KL.
The number of components of a basic polyhedron (i.e., its corresponding
alternating KL) is not the property of a single basic polyhedron, but of the
family of basic polyhedra and depends from family parameters.
The same holds for some other KL invariants, like signature or BJ-
unknotting (unlinking) number. This can be illustrated by the example of
the one-parameter family of basic polyhedra |2| sk0 s (k ≥ 3, s = 3 2 1 2),
starting with the basic polyhedron 9∗ . For k = 3, . . . , 21 we obtained the
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
302 LinKnot
Fig. 2.77 Composition of 3-tangles and Borromean rings obtained as its closure.
following table, where the second column gives the ordering number of the
basic polyhedron from the LinKnot data base (for k ≤ 14), the third column
the number of components, the fourth column the signature, and the fifth
column BJ-unknotting (unlinking) number of the basic polyhedron (see
page 83).
k BP σ u
3 9∗ 2 2
4 10∗∗ 2 1
5 11∗ 1 0
6 12B 1 0
7 133∗ 3 1
8 148∗ 2 0
9 1510∗ 3 2
10 1625∗ 3 1
11 17455∗ 2 0
12 182675∗ 2 0
13 195031∗ 3 1
14 2031002∗ 3 0
15 3 2
16 4 1
17 3 0
18 3 0
19 4 1
20 4 0
21 4 2
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
2.8 KL tables
The first (and still the best program) for knot theory is Knotscape with the
tables of knots with n ≤ 16 crossings, giving possibility to compute their
various invariants (Alexander, Jones, HOMFLYPT and Kauffman polyno-
mials, hyperbolic invariants, signature, symmetries, etc.) There are several
sources on the Internet providing knot data bases, containing polynomials
of knots and some other knot invariants. Almost all of them are based
on Hoste-Thistlethwaite Knotscape tables of knots (using only a part of
them: knots with n ≤ 11 or n ≤ 12 crossings). The best available source of
that kind is the Table of Knot Invariants by C. Livingston and J.C. Cha
(http://www.indiana.edu/ knotinfo/) which contains knots with n ≤ 12
crossings (and not links), providing a reader the most complete compu-
tational results for various knot invariants and some possibilities for in-
teractive computation. More possibilities for interactive use provides Knot
Atlas by Dror Bar Natan (http://katlas.math.toronto.edu/wiki/). Knot At-
las contains the tables of knots with n ≤ 11 crossings and links with n ≤ 11
crossings, which can be used for further computations in the Mathematica-
based program Knot Theory. Excellent program is R. Scharein’s Knot-
Plot (http://www.pims.math.ca/knotplot/), which has a superb graph-
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
304 LinKnot
KLs can be obtained from each generating link by adding 2 to the chains
of bigons. For example, from the generating rational link 3 1 2 with n = 6
crossings we obtain 5 1 2 and 3 1 4 for n = 8, then 7 1 2, 5 1 4, and 3 1
6 for n = 10, etc. As we pointed out many times, all important properties
of KLs (including their Dowker codes, polynomial invariants, minimum
braids, signatures, unlinking numbers, linking numbers, symmetry proper-
ties, etc.) are well ordered according to families.
Unfortunately, the data base KnotLinkBase.nb is perhaps the weakest
part of the program LinKnot, since it is created manually and probably
contains some errors or misprints.
Recall that the most of knot tables that are currently in use are based
on the more then 100 year old results of T.P. Kirkman (1885a,b), P.G. Tait
(1876/77a,b,c, 1883/84, 1884/85), and C.N. Little (1885, 1890, 1892, 1900),
and have been corrected several times. Computer derivations of KLs have
only appeared in the last few decades and are mostly restricted to knots
(Thistlethwaite, 1999), and links with a low number of crossings given in
Dowker notation (Doll and Hoste, 1991; Cerf, 1998). Based on recent re-
sults obtained by the program LinKnot, it is reasonable to expect that soon
it will be possible to derive KLs in Conway notation by computer, or at
least alternating KLs. We are able to generate all rational and stellar KLs
and all alternating KLs obtained from some source link– the only restric-
tion on the number of crossings is coming from computer limitations. As
it was already explained, derivation of alternating KLs is basically a series
of tangle substitutions made in previously generated source KLs, mostly
based on specific partitions or compositions of numbers. Symmetry plays
an important role in reducing the number of possibilities and recognizing
in advance possible repetitions and duplicates of KLs. Since we have no
general algorithmic solution for implementation of symmetry and its numer-
ous particular cases in a computer program for KL derivation, sometimes
is necessary to create all possible Conway symbols, and then select those
with different minimal Dowker codes computed by the LinKnot function
MinDowAltKL.
Non-alternating KLs of the polyhedral world represent a bigger prob-
lem, because the same non-alternating KL can be generated from different
basic polyhedra. For example, the Conway symbols 8∗ 2 : −2 0 and 9∗ − 2 0
represent the same non-alternating knot (Fig. 2.78). We can derive non-
alternating KLs given by Conway symbols, using the following algorithm:
306 LinKnot
(1) First filter contains some reduction procedures used in the initial gen-
eration of KLs;
(2) Polynomials are used at the next stage – they can be quickly computed,
although they are not very powerful at distinguishing KLs. They divide
KLs into relatively small equivalence classes;
(3) Now we introduce more powerful invariants, like homomorphisms of
the KL group or those arising from hyperbolic structure (if it exists).
Finally, one has to work hard to show that the groups of diagrams that
haven’t been distinguished so far, actually do represent the same KL.
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
For all KLs with n ≤ 11 crossings, with the exception of four of them,
we succeeded in finding their minimal almost alternating representations.
In their corresponding Conway symbols the sign +− denotes the crossing
change from a1 , ..., an + to a1 , ..., an −, and 1− denotes the crossing change
from +1 to −1.
Generally, the problem of finding minimal almost alternating represen-
tations is a difficult problem with a lot of open questions.
The minimality of almost alternating KL-representations can be proven
using graph-theoretical argumentation (Caudron, 1982), but for most of
them we have only empirical tools: construct all almost alternating repre-
sentations with a fixed number of crossings and choose the first correspond-
ing to a given non-alternating KL. Whenever it was possible, we requested
that the minimal almost alternating representation has to belong to the
same family (class, subworld,. . .) of the non-alternating KL.
For n ≤ 9 all the sources (Conway, 1970; Caudron, 1982; Adams, 1994;
Rolfsen, 1976) agree on the number of non-alternating knots: there are 3
non-alternating knots for n = 8, and 8 for n = 9. For n = 10 in Conway
(1970) and Rolfsen (1976) one knot was repeated (Perko pair, 10162 =
10161 ), i.e.,
2 1 : −2 0 : −2 0 = 3 : −2 0 : −2 0
After this correction, for n = 10 there are 42 non-alternating knots. For n =
11 crossings, 182 non-alternating knots are given in Conway’s paper, where
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
308 LinKnot
310 LinKnot
In the same way, from the following 13-crossing minimal almost alter-
nating representations we obtain the next six non-alternating knots:
. − (3, 2).2 .(3 1− , 2 1).2 . − (2 1, 2).2 .(3 1− , 3).2
.2. − (3, 2) .2.(3 1− , 2 1) .2. − (2 1, 2) .2.(3 1− , 3)
. − (3, 2).2 0 .(3 1− , 2 1).2 0 .2 0. − (3, 2) .2 0.(3 1− , 2 1)
ist, they must have more then 16 crossings. Knowing one of them, the
other follows immediately. For one of the knots from Adams’ question,
9∗ . − 2 : . − 2, we obtained 12-crossing minimal non-alternating representa-
tion 8∗ 2 0.2 0.1−.2 0.2 0.
A knot or link can have more then one minimal diagram. Every minimal
diagram of an alternating KL can be obtained from any other minimal
diagram of the same KL by a finite series of flypes (Theorem 1.11). If
a bigon is denoted by a bold line, an elementary flype can be illustrated
as in Fig. 2.80b. Flype diagrams (or vertex-bicolored diagrams) of KLs
are obtained in the following way: after collapsing every chain of bigons
with both ends incident to the same vertex into a black point, new bigons
obtained by the collapse are denoted by bold lines. In vertex-bicolored
diagrams, transition from one projection to another by a flype is represented
by a mutual place exchange of differently colored vertices connected by a
bold edge. For example, the transition from one minimal projection of the
link 2 2 2 to the other projection ((1, 2, 1), 1, 1) of the same link, expressed
in the language of flype diagrams, is illustrated in Fig. 2.80b. This basic
flyping algorithm can be applied only to algebraic tangles. Non-isomorphic
projections of KLs with n ≤ 9 crossings and their corresponding flype
diagrams are given in Figs. 2.81 and 2.82.
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
312 LinKnot
Fig. 2.80 (a) A flype; (b) an elementary flype and its vertex-colored interpretation; (c)
flype diagrams of the link 2 2 2.
Fig. 2.81 Projections of algebraic alternating source KLs with n ≤ 8 crossings and
their corresponding vertex-bicolored flype diagrams.
t 1 −1 0
0 −t 1 −1
dD (t) = = t4 − 2t2 .
−1 0 t 1
1 −1 0 −t
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
314 LinKnot
Fig. 2.82 Projections of polyhedral source KLs with n ≤ 9 crossings and their corre-
sponding vertex-bicolored flype diagrams.
Theorem 2.15. If dD′ (t) 6= dD′′ (t), two oriented alternating knot diagrams
D′ and D′′ are non-isomorphic (Jablan, 1995).
Proof. Let D′ and D′′ denote two isomorphic alternating knot diagrams,
and A′ and A′′ their corresponding matrices. If D′ and D′′ are isomorphic,
there exists a permutation matrix P such that P A′ P −1 = A′′ . From the re-
lations: |P A′ P −1 | = |A′′ |, |P A′ P −1 | = |P ||A′ ||P −1 | = |P ||A′ | |P1 | = |A′ | =
|A′′ |, |A′ | = |A′′ |, it follows that dD (t) = det(aij ) is the invariant of alter-
nating knot diagrams.
(1) for every alternating knot projection D, the degree of dD (t) is n, |cn | =
1 and |c1 | = |w(D)|, where w(D) is the writhe of D;
(2) dD (t) and dD (−t) correspond to obverse (mirror-symmetric) knot dia-
grams;
(3) for n = 0 (mod 2), a change of the orientation results in a change
of dD (t) to dD (−t), and for n = 1 (mod 2) in a change of dD (t) to
−dD (−t).
316 LinKnot
31 3 462 1 0 3
41 22 4682 1 0 −2 0
51 5 6 8 10 2 4 1 0 5 0 5
52 32 4 8 10 2 6 1 0 1 0 5
61 42 4 8 12 10 2 6 1 0 −4 −2 3 2
62 312 4 8 10 12 2 6 1 0 1 −1 −3 2
63 2112 4 8 10 2 12 6 1 0 2 0 3 0
71 7 8 10 12 14 2 4 6 1 0 7 0 14 0 7
72 52 4 10 14 12 2 8 6 1 0 −3 0 −2 0 7
73 43 6 10 12 14 2 4 8 −1 0 −1 0 8 0 7
74 313 6 10 12 14 4 2 8 1 0 −4 0 2 0 7
75 322 4 10 14 12 2 6 8 −1 0 −2 1 4 0 7
7′5 ((1, 3, 1), 1, 1) 4 10 12 14 2 8 6 −1 0 −3 0 4 0 7
76 2212 4 8 12 2 14 6 10 1 0 −1 −1 −1 −2 3
7′6 (((1, 2, 1), 1), 1, 1) 4 8 12 10 2 14 6 −1 0 0 −1 3 −1 3
77 21112 4 8 10 12 2 14 6 1 0 −1 −1 −1 −2 3
7′7 ((((1, 2), 1), 1), 1, 1) 4 8 12 14 2 6 10 −1 0 0 −1 3 −1 3
3, 2 1, −2
{{8}, {6, 8, 12, 2, −14, −16, 4, −10}}
{{8}, {4, 8, 12, 2, −14, 6, −16, −10}}
{{8}, {4, 8, 14, 2, −12, −16, 6, −10}}
{{8}, {4, 8, −12, 2, 14, 16, −6, 10}}
{{8}, {−6, 8, 12, −14, 4, 16, −2, 10}}
{{8}, {6, 8, 14, 12, 4, −16, 2, −10}}
{{8}, {6, 8, −10, 12, 14, −16, 2, −4}}
3, 3, −2
{{8}, {6, 8, −12, 2, 14, 16, −4, 10}}
{{8}, {−4, −8, 12, −2, 14, 16, 6, 10}}
{{8}, {6, −8, −12, 14, −4, 16, −2, 10}}
{{8}, {−6, 8, −14, 12, 4, 16, −2, 10}}
{{8}, {6, −8, 10, −12, 14, −16, 2, −4}}
318 LinKnot
Fig. 2.83 Achiral knots with n = 12 crossings from the work of M.G. Haseman.
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
In the case of the P -world for n ≤ 12, we will restrict the discussion of
achirality to the basic polyhedra and alternating KLs generated from them
as families. We conjecture that alternating achiral KLs can only be derived
from achiral basic polyhedra or achiral source KLs by an arrangement of
tangles preserving achirality of a generating KL.
We already mentioned that some more sensitive polynomial invariants
(e.g., Jones, Kauffman or HOMFLYPT polynomials) are able to recognize
chiral KLs, but not always.
The origin of polyhedral achiral knots will be achiral basic polyhedra
with a symmetry group G which contains rotational antireflection, antiro-
tation of order 2, or anti-inversion. The first achiral basic polyhedron is
6∗ (Borromean rings). For n ≤ 12 we obtain from it the following achiral
knots:
n=8
.2.2
n = 10
.2 1.2 1 .2.2.2 0.2 0 2.2.2.2
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
320 LinKnot
n = 12
.4.4 .3 1.3 1 .2 2.2 2 .3.3.2.2 .2 1.2 1.2.2
.3.3.2 0.2 0 .2 1.2.2 1 0.2 0 3.2.2.3 2.2 1.2 1.2 3 0.2.2.3 0
2 0.2 1.2 1.2 0 2.2.2.2.2 0.2 0 2.2.2 0.2.2.2 0
For 6∗ , G = [3, 4] and G′ = [3+ , 4], so the achiral knots derived from it
will contain antirotation or rotational antireflection. In this way, from 6∗
we derive non-invertible achiral knots (e.g., .2.2, .2.2.2.2, etc.), as well as
invertible ones (e.g., .2.2.2 0.2 0).
The invertible achiral knot 8∗ with the antireflection corresponds to the
basic polyhedron 8∗ . The following achiral knots are derived from the basic
polyhedron 8∗ :
n = 10
8∗ 2 0.2 0 8∗ 2 : .2
n = 12
8∗ 3.3 8∗ 2 1.2 1 8∗ 3 0.3 0 8∗ 3 : .3 8∗ 3 0 : .3 0
8∗ 2 1 0 : .2 1 0 8∗ .20 : 2.2 : 20 8∗ 2.2.2.2 8∗ 20.20.20.20 8∗ 2.20.20.2
8∗ 2.2 : .2 0.2 0 8∗ 2 0.2 0 : .2 0.2 0
Some of the obtained achirals are invertible (e.g. 8∗ , etc.), and others
are not (e.g., 8∗ 2 0.2 0, 8∗ 2 : .2, etc.).
The achiral invertible knot 10∗ corresponds to the basic polyhedron 10∗ .
The following achiral knots with n = 12 crossings are derived from 10∗ :
10∗ 2 ::: .2, 10∗ 2 0 ::: 2 0, 10∗ 2 :::: .2
From the basic polyhedron 10∗∗ we derive achiral knots:
n = 12
10∗∗ 2 :: .2 10∗∗ 2 0 :: .2 0 10∗∗ : 2.2 10∗∗ : 2 0.2 0 10∗∗ : 2 :: .2
10∗∗ : 2 0 :: .2 0
Fig. 2.84 Achiral knot H59 = H60 and two its vertex-antisymmetric 3D-presentations,
based on the anti-inversion.
322 LinKnot
Fig. 2.85 The antisymmetric presentation of the achiral knot (2 1, 3) 1 1 (3, 2 1).
links ((2m), (2n) + (2p)) ((2m), (2n) + 2p), (m, n, p ≥ 1); 3-component achi-
ral link .3.3 that generates the family of achiral links .(2m + 1).(2m + 1),
(m ≥ 1); 3-component achiral link .3 : 30 that generates the family of achi-
ral links .(2m + 1) : (2m + 1) 0; and 3-component achiral link 8∗ 2.2 that
generates the family of achiral links 8∗ (2m + 1).(2m + 1), (m ≥ 1). For
n = 12, together with the links belonging to the families already mentioned
(4, 4, 2, 2; 4, 2, 4, 2; (4, 2) (4, 2); (2, 2 + +) (2, 2 + +), and 4. : 4 0), we have
the first elements of new families of achiral links:
and two achiral basic polyhedra 121∗ (12A) and 1210∗ (12J).
In the same way, we can try to find achiral non-alternating KLs. It is
interesting that in the existing knot tables for n ≤ 10 there are no non-
alternating achiral knot. The first non-alternating achiral knot 1211∗ − 1. −
1. − 1 :: −1. − 1. − 1 has n = 12 crossings (Fig. 2.86).
We give the list of achiral non-alternating knots with n = 14 crossings
and their corresponding families, and the analogous lists of achiral non-
alternating links with n = 10 crossings, but we can not guarantee that our
lists are complete.
For n = 14 we have the following non-alternating achiral knots and the
families of achiral knots generated from them4 :
324 LinKnot
326 LinKnot
alternating KLs originate from achiral basic polyhedra and can be obtained
by symmetric tangle substitutions.
Every palindromic rational knot with an even number of crossings is
achiral, and the same statement holds for rational links (Theorem 1.20,
Theorem 1.22). We hope that the more general statement holds: every
algebraic alternating KL of the symmetric (palindromic) form p p is achiral,
where p is any algebraic tangle, and we suppose that all achiral alternating
algebraic KLs can be described in this way.
For achiral alternating polyhedral KLs, origins are achiral basic poly-
hedra (Fig. 2.88). Their list for n ≤ 16 is:
n=6
6∗
No. of basic polyhedra: 1
n=8
8∗
No. of basic polyhedra: 1
n = 10
10∗ 10∗∗ 10∗∗∗
No. of basic polyhedra: 3
n = 12
121∗ 122∗ 127∗ 128∗ 1210∗ 1211∗ 1212∗
No. of basic polyhedra: 7
n = 14
144∗ 147∗ 148∗ 149∗ 1411∗ 1412∗ 1413∗ 1417∗
1419∗ 1420∗ 1428∗ 1433∗ 1444∗ 1447∗ 1449∗ 1450∗
1453∗ 1455∗ 1456∗ 1458∗ 1460∗
No. of basic polyhedra: 21
n = 16
162∗ 167∗ 1617∗ 1621∗ 1623∗ 1625∗ 1632∗ 1639∗
1643∗ 1646∗ 1647∗ 1648∗ 1649∗ 1651∗ 1674∗ 1680∗
1685∗ 1686∗ 1689∗ 1692 ∗ 1697∗ 16110∗ 16113∗ 16128∗
16132∗ 16133∗ 16142∗ 16150∗ 16156∗ 16160∗ 16175∗ 16204∗
16206∗ 16223∗ 16226∗ 16227∗ 16230∗ 16234∗ 16235∗ 16239∗
16242∗ 16256∗ 16263∗ 16270∗ 16273∗ 16280∗ 16282∗ 16284∗
16285∗ 16286∗ 16293∗ 16327∗ 16346∗ 16347∗ 16351∗ 16360∗
16361∗ 16367∗ 16369∗ 16374∗ 16377∗ 16380∗ 16384∗ 16388∗
16391∗ 16393∗ 16402∗ 16412∗ 16416∗ 16419∗ 16428∗ 16429∗
16431∗ 16437∗ 16442∗ 16451 ∗ 16461∗ 16494∗
No. of basic polyhedra: 78
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
6∗ p.p 6∗ p : .p
6∗ p.q.p 0:q 6∗ p.p : q 0.q 0 6∗ p.p : q.q
6∗ p.q : p.q 6∗ p.q.q.p
6∗ p.p.q 0.r.r.q 0 6∗ p.q.r 0.p.q.r 0 6∗ p.q.q.p.r 0.r 0
6∗ p.q.r.p.r 0.q 0 6∗ p.p.q.r.r.q 6∗ p.q.r.p.q.r
328 LinKnot
8∗ p : .p 8∗ p 0.p 0 8∗ p 0 : .p 0
8∗ p.p
8∗ p.q 0.q 0.p : r 0.r 0 8∗ p.p.q 0 : r 0.r 0 : q 0 8∗ p 0.q 0.q 0.p 0 : r 0.r 0
8∗ p.p : q.r 0.r 0.q 8∗ p.p.q 0.r : .r.q 0 8∗ p.q 0.r 0.r 0.q 0.p
8∗ p.q.r.r.q.p 8∗ p.q.q.p : r.r 8∗ p.p.q 0.r 0 : .r 0.q 0
8∗ p 0.q 0.r 0.r 0.q 0.p 0 8∗ p.q.q.p.r 0 : .r 0 8∗ p.p : q 0.r 0.r 0.q 0
8∗ p.q.r 0.r 0.q.p 8∗ p.p.q 0 : r.r : q 0 8∗ p.q.q.p : r 0.r 0
8∗ p.q 0.q 0.p.r 0 : .r 0
8∗ p.p.q 0.r 0.s 0.s 0.r 0.q 0 8∗ p.p.q 0.r 0.s.s.r 0.q 0 8∗ p.q.r 0.r 0.q.p.s 0.s 0
8∗ p.q.q.p.r 0.s 0.s 0.r 0 8∗ p.q.r.r.q.p.s 0.s 0 8∗ p.q.q.p.r 0.s.s.r 0
8∗ p 0.p 0.q 0.r 0.s 0.s 0.r 0.q 0 8∗ p.q 0.q 0.p.r 0.s 0.s 0.r 0 8∗ p.p.q 0.r.s 0.s 0.r.q 0
8∗ p.p.q.r.s.s.r.q
From the basic polyhedron 10∗ the following classes of achiral alternat-
ing KLs are derived:
10∗ p 0.p 0 : q 0 : r 0.r 0 : q 0 10∗ p 0.q 0 : r 0 : p 0.q 0 : r 0 10∗ p 0.q 0 : r 0.r 0 : q 0.p 0
10∗ p 0.q 0.r 0 : .p 0.q 0.r 0 10∗ p 0.q 0.r 0 : .r 0.q 0.p 0 10∗ p 0.q 0.q 0.p 0 : .r 0.r 0
10∗ p 0.q 0.q 0.p 0 : r 0 : .r 0 10∗ p 0.q 0.r 0.r 0.q 0.p 0 10∗ p : q 0.r 0.r 0.q 0 : p
10∗ p.q 0 : r 0.r 0 : q 0.p 10∗ p.q 0.r 0 : .r 0.q 0.p 10∗ p.q 0.r 0.r 0.q 0.p
10∗ p : q 0.q 0 : p : r 0.r 0 10∗ p : q 0.r 0 : p : q 0.r 0 10∗ p.q 0 : .q 0.p : r 0.r 0
10∗ p.q 0 : .q 0.p.r 0 : .r 0 10∗ p.q 0 : .r 0.p.q 0 : .r 0 10∗ p.q 0 : r 0 : p.q 0 : r 0
10∗ p.q 0.r 0 : .p.q 0.r 0 10∗ p.q 0.q 0.p : .r 0.r 0 10∗ p.q 0.q 0.p : r 0 : .r 0
10∗ p.q 0.q 0.p.r 0 :: .r 0 10∗ p.q 0.q 0.p : r : .r 10∗ p.q 0 : r : p.q 0 : r
10∗ p.q 0.r : .p.q 0.r 10∗ p.q 0.r : .r.q 0.p 10∗ p.p : .q 0.r 0.r 0.q 0
10∗ p.p : q 0 : r 0.r 0 : q 0 10∗ p.p : q 0.r 0 : .r 0.q 0 10∗ p.p.q 0 : .r 0.r 0 : .q 0
10∗ p.p.q 0 : r 0 : .r 0 : q 0 10∗ p.p.q 0.r 0 :: .r 0.q 0 10∗ p.q : r 0.r 0 : q.p
10∗ p.q.r 0 : .r 0.q.p 10∗ p.q.r 0.r 0.q.p 10∗ p.p : q 0 : r.r : q 0
10∗ p.q : r 0 : p.q : r 0 10∗ p.p : q 0.r : .r.q 0 10∗ p.p.q 0 : .r.r : .q 0
10∗ p.q.r 0 : .p.q.r 0 10∗ p.p.q 0 : r : .r : q 0 10∗ p.p.q 0.r :: .r.q 0
10∗ p.p : .q.r 0.r 0.q 10∗ p.p : q : r 0.r 0 : q 10∗ p.p : q.r 0 : .r 0.q
10∗ p.q.q.p : .r 0.r 0 10∗ p.q.q.p : r 0 : .r 0 10∗ p.q.q.p.r 0 :: .r 0
10∗ p.p : q : r.r : q 10∗ p.q : r : p.q : r 10∗ p.q : r.r : q.p
10∗ p.q.r : .p.q.r 10∗ p.q.r : .r.q.p 10∗ p.q.q.p : .r.r
10∗ p.q.q.p : r : .r 10∗ p.q.r.r.q.p
10∗ p 0.q 0.q 0.p 0 : r 0.s 0.s 0.r 0 10∗ p 0.q 0.r 0.s 0 : p 0.q 0.r 0.s 0
10∗ p 0.q 0.r 0.r 0.q 0.p 0 : s 0.s 0 10∗ p 0.q 0.r 0.s 0.s 0.r 0.q 0.p 0
10∗ p.q 0.r 0.s 0.s 0.r 0.q 0.p 10∗ p.q 0.r 0.r 0.q 0.p : s 0.s 0
10∗ p.q 0.r 0.r 0.q 0.p.s 0 : .s 0 10∗ p.q 0.r 0 : s 0.p.q 0.r 0 : s 0
10∗ p.q 0.r 0.s 0 : p.q 0.r 0.s 0 10∗ p.q 0.q 0.p : r 0.s 0.s 0.r 0
10∗ p.q 0.q 0.p.r 0 : s 0.s 0 : r 0 10∗ p.q 0.q 0.p.r 0.s 0 : .s 0.r 0
10∗ p.q 0.q 0.p : r.s 0.s 0.r 10∗ p.q 0.r 0.s : p.q 0.r 0.s
10∗ p.q 0.r.s 0.s 0.r.q 0.p 10∗ p.q 0.r.s 0 : p.q 0.r.s 0
10∗ p.p : q 0.r 0.s 0.s 0.r 0.q 0 10∗ p.p.q 0 : r 0.s 0.s 0.r 0 : q 0
10∗ p.p.q 0.r 0 : s 0.s 0 : r 0.q 0 10∗ p.p.q 0.r 0.s 0 : .s 0.r 0.q 0
10∗ p.q.r 0.s 0.s 0.r 0.q.p 10∗ p.p.q 0 : r 0.s.s.r 0 : q 0
10∗ p.q.r 0 : s 0.p.q.r 0 : s 0 10∗ p.p.q 0.r 0 : s.s : r 0.q 0
10∗ p.q.r 0.s 0 : p.q.r 0.s 0 10∗ p.p.q 0.r 0.s : .s.r 0.q 0
10∗ p.q.r 0.r 0.q.p : s 0.s 0 10∗ p.q.r 0.r 0.q.p.s 0 : .s 0
10∗ p.q.r 0.r 0.q.p : s.s 10∗ p.p : q 0.r.s 0.s 0.r.q 0
10∗ p.p.q 0 : r.s 0.s 0.r : q 0 10∗ p.p.q 0.r : s 0.s 0 : r.q 0
10∗ p.p.q 0.r.s 0 : .s 0.r.q 0 10∗ p.q.r 0.s.s.r 0.q.p
10∗ p.q.r 0.s : p.q.r 0.s 10∗ p.p.q 0.r : s.s : r.q 0
10∗ p.p : q.r 0.s 0.s 0.r 0.q 10∗ p.q.r.s 0.s 0.r.q.p
10∗ p.q.r.s 0 : p.q.r.s 0 ∗
10 p.q.q.p : r 0.s 0.r 0.r 0
10∗ p.q.q.p.r 0 : s 0.s 0 : r 0 10∗ p.q.q.p.r 0.s 0 : .s 0.r 0
10∗ p.q.q.p : r 0.s.s.r 0 10∗ p.q.q.p.r 0 : s.s : r 0
10∗ p.q.q.p.r 0.s : .s.r 0 10∗ p.q.q.p : r.s 0.s 0.r
10∗ p.q.r.r.q.p : s 0.s 0 10∗ p.q.r.r.q.p.s 0 : .s 0
10∗ p.q.q.p : r.s.s.r 10∗ p.q.r.s : p.q.r.s
10∗ p.q.q.p.r.s : s.r 10∗ p.q.r.s.s.r.q.p
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
330 LinKnot
10∗ p 0.p 0.q 0.r 0.s 0.t 0.t 0.s 0.r 0.q 0 10∗ p 0.q 0.q 0.p 0.r 0.s 0.t 0.t 0.s 0.r 0
10∗ p 0.q 0.r 0.r 0.q 0.p 0.s 0.t 0.t 0.s 0 10∗ p 0.q 0.r 0.s 0.t 0.p 0.q 0.r 0.s 0.t 0
10∗ p.q 0.r 0.r 0.q 0.p.s 0.t 0.t 0.s 0 10∗ p.q 0.r 0.s 0.t 0.p.q 0.r 0.s 0.t 0
10∗ p.q 0.q 0.p.r 0.s 0.t 0.t 0.s 0.r 0 10∗ p.q 0.r.s 0.t 0.p.q 0.r.s 0.t 0
10∗ p.p.q 0.r 0.s 0.t 0.t 0.s 0.r 0.q 0 10∗ p.p.q 0.r 0.s 0.t.t.s 0.r 0.q 0
10∗ p.q.r 0.s 0.t 0.p.q.r 0.s 0.t 0 10∗ p.p.q 0.r 0.s.t 0.t 0.s.r 0.q 0
10∗ p.q.r 0.r 0.q.p.s 0.t 0.t 0.s 0 10∗ p.p.q 0.r.s 0.t 0.t 0.s 0.r.q 0
10∗ p.p.q 0.r.s 0.t.t.s 0.r.q 0 10∗ p.q.r 0.s.t 0.p.q.r 0.s.t 0
10∗ p.q.r 0.s.s.r 0.q.p.t 0.t 0 10∗ p.q.r.s 0.t 0.p.q.r.s 0.t 0
10∗ p.q.q.p.r 0.s 0.t 0.t 0.s 0.r 0 10∗ p.q.q.p.r 0.s 0.t.t.s 0.r 0
10∗ p.q.q.p.r 0.s.t 0.t 0.s.r 0 10∗ p.q.q.p.r 0.s.t.t.s.r 0
10∗ p.q.r.s.t 0.p.q.r.s.t 0 10∗ p.q.r.r.q.p.s 0.t 0.t 0.s 0
10∗ p.q.r.r.q.p.s 0.t.t.s 0 10∗ p.q.r.s.s.r.q.p.t 0.t 0
10∗ p.p.q.r.s.t.t.s.r.q 10∗ p.q.q.p.r.s.t.t.s.r
10∗ p.q.r.r.q.p.s.t.t.s 10∗ p.q.r.s.t.p.q.r.s.t
From the basic polyhedron 10∗∗ the following classes of achiral alternat-
ing KLs are derived:
10∗∗ p 0.q 0 : r : p 0.q 0 : r 10∗∗ p 0.q 0.r : .p 0.q 0.r 10∗∗ p 0.q 0.r.r.q 0.p 0
10∗∗ .p 0.q : r 0 : p 0.q : r 0 10∗∗ .p 0.q.r 0 : .p 0.q.r 0 10∗∗ p 0 : q.r 0 : p 0 : q.r 0
10∗∗ p 0.q : .r 0.p 0.q : .r 0 10∗∗ p 0.q : r 0 : p 0.q : r 0 10∗∗ p 0.q.r 0 : .p 0.q.r 0
10∗∗ p 0.q.r 0.r 0.q.p 0 10∗∗ .p 0.q.q.p 0 : .r 0.r 0 10∗∗ .p 0.q.q.p 0 : r 0 : .r 0
10∗∗ p 0 : q.q : p 0 : r 0.r 0 10∗∗ p 0.q : .q.p 0 : r 0.r 0 10∗∗ p 0.q : .q.p 0.r 0 : .r 0
10∗∗ .p 0.q.r : .p 0.q.r 10∗∗ .p 0.q.q.p 0 : .r.r 10∗∗ p 0 : q.q : p 0 : r.r
10∗∗ p 0 : q.r : p 0 : q.r 10∗∗ p 0.q : .q.p 0 : r.r 10∗∗ p 0.q : .q.p 0.r : .r
10∗∗ p 0.q : .r.p 0.q : .r 10∗∗ p 0.q : r : p 0.q : r 10∗∗ p 0.q.r : .p 0.q.r
10∗∗ p 0.q.r.r.q.p 0 10∗∗ p.q 0.r 0.r 0.q 0.p ∗∗
10 .p : q 0.r 0 : p : q 0.r 0
10∗∗ .p.q 0 : r 0 : p.q 0 : r 0 10∗∗ .p.q 0.r 0 : .p.q 0.r 0 10∗∗ .p.q 0.q 0.p : .r 0.r 0
10∗∗ .p.q 0.q 0.p : r 0 : .r 0 10∗∗ p : q 0.q 0 : p : r 0.r 0 10∗∗ p : q 0.r 0 : p : q 0.r 0
10∗∗ p.q 0 : .q 0.p : r 0.r 0 10∗∗ p.q 0 : .q 0.p.r 0 : .r 0 10∗∗ p.q 0 : .r 0.p.q 0 : .r 0
10∗∗ p.q 0 : r 0 : p.q 0 : r 0 10∗∗ p.q 0.r 0 : .p.q 0.r 0 10∗∗ .p.q 0.q 0.p : .r.r
10∗∗ .p.q 0.q 0.p : r : .r 10∗∗ p.q 0 : .q 0.p : r.r 10∗∗ .p.q 0 : r : p.q 0 : r
10∗∗ .p.q 0.r : .p.q 0.r 10∗∗ p.q 0 : r : p.q 0 : r 10∗∗ p.q 0.r : .p.q 0.r
10∗∗ p.q 0.r.r.q 0.p 10∗∗ .p : .p : q 0.r 0.r 0.q 0 10∗∗ p.q.r 0.r 0.q.p
10∗∗ .p : .p : q 0.r.r.q 0 10∗∗ .p : q.r 0 : p : q.r 0 10∗∗ .p.q : r 0 : p.q : r 0
10∗∗ .p.q.r 0 : .p.q.r 0 10∗∗ p : q.r 0 : p : q.r 0 10∗∗ p.q : .r 0.p.q : .r 0
10∗∗ p.q : r 0 : p.q : r 0 10∗∗ p.q.r 0 : .p.q.r 0 10∗∗ .p.q.q.p : .r 0.r 0
10∗∗ .p.q.q.p : r 0 : .r 0 10∗∗ p : q.q : p : r 0.r 0 10∗∗ p.q : .q.p : r 0.r 0
10∗∗ p.q : .q.p.r 0 : .r 0 10∗∗ .p.q : r : p.q : r 10∗∗ .p.q.r : .p.q.r
10∗∗ .p.q.q.p : .r.r 10∗∗ .p.q.q.p : r : .r 10∗∗ p : q.q : p : r.r
10∗∗ p : q.r : p : q.r 10∗∗ p.q : .q.p : r.r 10∗∗ p.q : .q.p.q : .r
10∗∗ p.q : .r.p.q : .r 10∗∗ p.q : r : p.q : r 10∗∗ p.q.r : .p.q.r
10∗∗ p.q.r.r.q.p
10∗∗ .p 0.q 0.q 0.p 0 : r 0.s 0.s 0.r 0 10∗∗ .p 0.q 0.r 0.s 0 : p 0.q 0.r 0.s 0
10∗∗ p 0.q 0.r 0 : s 0.p 0.q 0.r 0 : s 0 10∗∗ p 0.q 0.r 0.s 0 : p 0.q 0.r 0.s 0
10∗∗ p 0.q 0.r 0.r 0.q 0.p 0 : s 0.s 0 10∗∗ p 0.q 0.r 0.r 0.q 0.p 0.s 0 : .s 0
10∗∗ p 0.q 0.r 0.r 0.q 0.p 0 : s.s 10∗∗ p 0.q 0.r 0.s : p 0.q 0.r 0.s
10∗∗ p 0.q 0.r : s 0.p 0.q 0.r : s 0 10∗∗ p 0.q 0.r.s 0 : p 0.q 0.r.s 0
10∗∗ p 0.q 0.r.r.q 0.p 0 : s 0.s 0 10∗∗ p 0.q 0.r.r.q 0.p 0.s 0 : .s 0
10∗∗ p 0.q 0.r.s : p 0.q 0.r.s 10∗∗ p 0.q 0.r.r.q 0.p 0 : s.s
10∗∗ .p 0.q.r 0.s 0 : p 0.q.r 0.s 0 10∗∗ p 0.q : r 0.s 0.p 0.q : r 0.s 0
10∗∗ p 0.q.r 0 : s 0.p 0.q.r 0 : s 0 10∗∗ p 0.q.r 0.s 0 : p 0.q.r 0.s 0
10∗∗ p 0.q.r 0.r 0.q.p 0 : s 0.s 0 10∗∗ p 0.q.r 0.r 0.q.p 0.s 0 : .s 0
10∗∗ p 0.q.r 0.r 0.q.p 0 : s.s 10∗∗ p 0.q.r 0.r 0.q.p 0.s : .s
10∗∗ p 0.q.r 0 : s.p 0.q.r 0 : s 10∗∗ p 0.q.r 0.s : p 0.q.r 0.s
10∗∗ .p 0.q.q.p 0 : r 0.s 0.s 0.r 0 10∗∗ p 0.q : .q.p 0.r 0.s 0.s 0.r 0
10∗∗ .p 0.q.q.p 0 : r 0.s.s.r 0 10∗∗ .p 0.q.r.s 0 : p 0.q.r.s 0
10∗∗ p 0.q : .q.p 0.r 0.s.s.r 0 10∗∗ p 0.q : r.s 0.p 0.q : r.s 0
10∗∗ p 0.q.r : s 0.p 0.q.r : s 0 10∗∗ p 0.q.r.s 0 : p 0.q.r.s 0
10∗∗ p 0.q.r : s.p 0.q.r : s 10∗∗ p 0.q.r.s : p 0.q.r.s
10∗∗ p 0.q.r.r.q.p 0 : s 0.s 0 10∗∗ p 0.q.r.r.q.p 0.s 0 : .s 0
10∗∗ p 0.q.r.r.q.p 0 : s.s 10∗∗ p 0.q.r.r.q.p 0.s : .s
10∗∗ p.q 0.r 0.r 0.q 0.p : s 0.s 0 10∗∗ p.q 0.r 0.r 0.q 0.p.s 0 : .s 0
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
332 LinKnot
10∗∗ p.q 0.r 0.r 0.q 0.p : s.s 10∗∗ .p.q 0.r 0.s 0 : p.q 0.r 0.s 0
10∗∗ p.q 0.r 0 : s 0.p.q 0.r 0 : s 0 10∗∗ p.q 0.r 0.s 0 : p.q 0.r 0.s 0
10∗∗ .p.q 0.q 0.p : r 0.s 0.s 0.r 0 10∗∗ .p.q 0.q 0.p : r 0.s.s.r 0
10∗∗ .p.q 0.q 0.p : r.s 0.s 0.r 10∗∗ .p.q 0.r 0.s : p.q 0.r 0.s
10∗∗ p.q 0.r 0.s : p.q 0.r 0.s 10∗∗ .p.q 0.r.s 0 : p.q 0.r.s 0
10∗∗ p.q 0.r : s 0.p.q 0.r : s 0 10∗∗ p.q 0.r.s 0 : p.q 0.r.s 0
10∗∗ p.q 0.r.r.q 0.p : s 0.s 0 10∗∗ p.q 0.r.r.q 0.p.s 0 : .s 0
10∗∗ p.q 0.r.r.q 0.p : s.s 10∗∗ p.q 0.r.s : p.q 0.r.s
10∗∗ .p.q.r 0.s 0 : p.q.r 0.s 0 ∗∗
10 p.q : r 0.s 0.p.q : r 0.s 0
10∗∗ p.q.r 0 : s 0.p.q.r 0 : s 0 10∗∗ p.q.r 0.s 0 : p.q.r 0.s 0
10∗∗ p.q.r 0.r 0.q.p : s 0.s 0 10∗∗ p.q.r 0.r 0.q.p.s 0 : .s 0
10∗∗ p.q.r 0.r 0.q.p : s.s 10∗∗ p.q.r 0.r 0.q.p.s : .s
10∗∗ .p.q.r 0.s : p.q.r 0.s 10∗∗ p.q.r 0 : s.p.q.r 0 : s
10∗∗ p.q.r 0.s : p.q.r 0.s 10∗∗ .p.q.r.s 0 : p.q.r.s 0
10∗∗ p.q : r.s 0.p.q : r.s 0 10∗∗ p.q.r : s 0.p.q.r : s 0
10∗∗ p.q.r.s 0 : p.q.r.s 0 10∗∗ .p.q.q.p : r 0.s 0.s 0.r 0
10∗∗ q.p : .p.q.r 0.s 0.s 0.r 0 10∗∗ .p.q.q.p : r 0.s.s.r 0
10∗∗ p.q : .q.p.r 0.s.s.r 0 10∗∗ .p.q.q.p : r.s 0.s 0.r
10∗∗ p.q.r.r.q.p : s 0.s 0 10∗∗ q.r.s.s.r.q.p 0 : .p 0
10∗∗ .p.q.q.p : r.s.s.r 10∗∗ p.q.r : s.p.q.r : s
10∗∗ p.q.r.s : p.q.r.s 10∗∗ p.q.q.p.r.s : s.r
10∗∗ p.q.r.r.q.p.s : .s
10∗∗ p 0.q 0.r 0.r 0.q 0.p 0.s 0.t 0.t 0.s 0 10∗∗ p 0.q 0.r 0.s 0.t 0.p 0.q 0.r 0.s 0.t 0
10∗∗ p 0.q 0.r.s 0.t 0.p 0.q 0.r.s 0.t 0 10∗∗ p 0.q 0.r.r.q 0.p 0.s 0.t 0.t 0.s 0
10∗∗ p 0.q 0.r.r.q 0.p 0.s 0.t.t.s 0 10∗∗ p 0.q 0.r.s.t 0.p 0.q 0.r.s.t 0
10∗∗ p 0.q.r 0.s 0.t 0.p 0.q.r 0.s 0.t 0 10∗∗ p 0.q.r 0.r 0.q.p 0.s 0.t 0.t 0.s 0
10∗∗ p 0.q.r 0.q 0.q.p 0.s 0.t.t.s 0 10∗∗ p 0.q.r 0.r 0.q.p 0.s.t 0.t 0.s
10∗∗ p 0.q.r 0.s 0.t.p 0.q.r 0.s 0.t 10∗∗ p 0.q.r 0.s.t 0.p 0.q.r 0.s.t 0
10∗∗ p 0.q.r.s 0.t 0.p 0.q.r.s 0.t 0 10∗∗ p 0.q.r.s 0.t.p 0.q.r.s 0.t
10∗∗ p 0.q.r.s.t 0.p 0.q.r.s.t 0 10∗∗ p 0.q.r.r.q.p 0.s 0.t 0.t 0.s 0
10∗∗ p 0.q.r.r.q.p 0.s 0.t.t.s 0 10∗∗ p 0.q.r.r.q.p 0.s.t 0.t 0.s
10∗∗ p 0.q.r.r.q.p 0.s.t.t.s 10∗∗ p 0.q.r.s.t.p 0.q.r.s.t
10∗∗ p.q 0.r 0.r 0.q 0.p.s 0.t 0.t 0.s 0 10∗∗ p.q 0.r 0.s 0.t 0.p.q 0.r 0.s 0.t 0
10∗∗ p.q 0.r.s 0.t 0.p.q 0.r.s 0.t 0 10∗∗ p.q 0.r.r.q 0.p.s 0.t 0.t 0.s 0
10∗∗ p.q 0.r.r.q 0.p.s 0.t.t.s 0 10∗∗ p.q 0.r.s.t 0.p.q 0.r.s.t 0
10∗∗ p.q.r 0.s 0.t 0.p.q.r 0.s 0.t 0 10∗∗ p.q.r 0.r 0.q.p.s 0.t 0.t 0.s 0
10∗∗ p.q.r 0.r 0.q.p.s 0.t.t.s 0 10∗∗ p.q.r 0.r 0.q.p.s.t 0.t 0.s
10∗∗ p.q.r 0.s 0.t.p.q.r 0.s 0.t 10∗∗ p.q.r 0.s.t 0.p.q.r 0.s.t 0
10∗∗ p.q.r.s 0.t 0.p.q.r.s 0.t 0 10∗∗ p.q.r.s 0.t.p.q.r.s 0.t
10∗∗ p.q.r.s.t 0.p.q.r.s.t 0 10∗∗ p.q.r.r.q.p.s 0.t 0.t 0.s 0
10∗∗ p.q.r.r.q.p.s 0.t.t.s 0 10∗∗ p.q.r.r.q.p.s.t 0.t 0.s
10∗∗ p.q.r.r.q.p.s.t.t.s 10∗∗ p.q.r.s.t.p.q.r.s.t
Fig. 2.89 The achiral basic polyhedron 10∗∗∗ expressed as the product of two non-
algebraic tangles 5⋆ .
First surprise comes as the result of of the chirality tests using Kauff-
man polynomial and the program SnapPea: 2-component link 10∗∗∗ 2 :: .2 0
recognized by Kauffman polynomials as achiral is chiral. All other polyno-
mials presumably able to detect chirality, Jones and HOMFLYPT polyno-
mial, fail as well. All chiral KLs that can not be recognized as chiral by any
of polynomials mentioned will be called non-detectable chiral KLs. More-
over, chiral link 10∗∗∗ 2 :: .2 0 generates an infinite class of non-detectable
chiral KLs of the form 10∗∗∗ p :: .p 0. The same property holds for chiral
source KLs 10∗∗∗ 2 :: .2 0 : .2 0.2, 10∗∗∗ 2 :: .2 0 : .2.2 0, 10∗∗∗ 2 : 2 0 : .2 : 2 0,
10∗∗∗ 2 : 2 : .2 0 : 2 0, 10∗∗∗ 2.2 : .2 0 : .2 0, 10∗∗∗ 2 : 2 0 : .2 : 2 0.2.2 0, etc.,
and infinite classes of chiral KLs 10∗∗∗ p :: .p 0 : .q 0.q, 10∗∗∗ p :: .p 0 : .q.q 0,
10∗∗∗ p : p 0 : .q : q 0, 10∗∗∗ p : q 0 : .q : p 0, 10∗∗∗ p.q 0 : .q : .p 0,
10∗∗∗ p : q : .p 0 : q 0, 10∗∗∗ p : q : .q 0 : p 0, 10∗∗∗ p.q : .q 0 : .p 0,
10∗∗∗ p : q 0 : .q : p 0.r.r 0, etc.
By different substitutions in the same achiral source KL we can obtain
classes of achiral KLs and non-detectable chiral KLs. For example, source
knot 10∗∗∗ 2 : .2 0.2.2 0 generates an infinite class of achiral KLs 10∗∗∗ p :
.q 0.p.q 0 and an infinite class of non-detectable chiral KLs 10∗∗∗ p : .q 0.q.p 0;
source knot 10∗∗∗ 2.2 0 : .2 : .2 0 generates an infinite class of achiral KLs
10∗∗∗ p.q 0 : .p : .q 0 and an infinite class of non-detectable chiral KLs
10∗∗∗ p.q 0 : .q : .p 0, etc.
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
334 LinKnot
Fig. 2.90 (a) Diagram of the achiral source knot 10∗∗∗ 2 : 2 : .2 0 : 2 0.2.2 0; (b) the
achiral knot 10∗∗∗ 2 : 2 : .2 0 : 2 0.2 1.2 1 0 without antisymmetric minimal projection.
Among classes of achiral KLs derived from source KLs with four bigons
there is one exceptional one-parameter class: the class of links 10∗∗∗ p.p.p.p.
A 2-parameter class 10∗∗∗ p : p : .p 0 : p 0.q.q 0 is the origin of a very
interesting subclass: class of knots with no minimal antisymmetric (i.e.,
achiral) diagram. For n ≤ 16 all achiral alternating knots have an achiral
minimal diagram. The first knot without it is 18-crossing achiral knot
10∗∗∗ 2 : 2 : .2 0 : 2 0.2 1.2 1 0. A simple explanation of this phenom-
enon is the following: source knot 10∗∗∗ 2 : 2 : .2 0 : 2 0.2.2 0 has an
antisymmetric minimal diagram (Figs. 2.90a, 2.91a). Its achirality is
the result of the rotational antireflection of order 4 (Fig. 2.91b). The
achiral knot 10∗∗∗ 2 : 2 : .2 0 : 2 0.2 1.2 1 0 has only one minimal dia-
gram, 10∗∗∗ 2 : 2 : .2 0 : 2 0.(2, 1).(2, 1) 0 which is chiral (Fig. 2.90b).
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
Fig. 2.91 (a) Achiral source knot 10∗∗∗ 2 : 2 : .2 0 : 2 0.2.2 0; (b) its 3-D antisymmetric
representation with the visible rotational antireflection of order 4.
336 LinKnot
Fig. 2.92 Dynamic antisymmetric 3-D representation of the achiral knot 10∗∗∗ 2 : 2 :
.2 0 : 2 0.2 1.2 1 0 with the rotational antireflection of order 4 preserved since tangles 2 1
and 1 2 are flype-equivalent.
10∗∗∗ p.q 0.r.s 0.p.s 0.r.q 0 10∗∗∗ p.q 0.r : r.q 0.p : s 0.s
10∗∗∗ p.q 0.r : r.q 0.p : s.s 0 10∗∗∗ p.q 0.r : p : r.q 0.s 0.s
10∗∗∗ p.q 0.r : p : r.q 0.s.s 0 10∗∗∗ p.q.r 0.s 0.p.s 0.r 0.q
10∗∗∗ p.q.r 0 : r 0.q.p : s 0.s 10∗∗∗ p.q.r 0 : r 0.q.p : s.s 0
10∗∗∗ p.q : r 0.p.r 0 : q.s 0.s 10∗∗∗ p.q.r 0 : p : r 0.q.s 0.s
10∗∗∗ p.q : r 0.p.r 0 : q.s.s 0 10∗∗∗ p.q.r 0 : p : r 0.q.s.s 0
10∗∗∗ p.q : r 0 : q.p.r 0.s 0.s 10∗∗∗ p.q : r 0 : q.p.r 0.s.s 0
10∗∗∗ p.q.r.s 0.p.s 0.r.q 10∗∗∗ p.q.r.s 0.r.q.p.s 0
10∗∗∗ p.q.r : p : r.q.s 0.s 10∗∗∗ p.q.r : r.q.p : s 0.s
10∗∗∗ p.q.r.s.p.s.r.q 10∗∗∗ p.q.r : p : r.q.s.s 0
10∗∗∗ p.q.r : r.q.p : s.s 0
10∗∗∗ p.q 0.r.s 0.p.s 0.r.q 0.t 0.t 10∗∗∗ p.q 0.r.s 0.p.s 0.r.q 0.t.t 0
10∗∗∗ p.q.r 0.s 0.p.s 0.r 0.q.t 0.t 10∗∗∗ p.q.r 0.s 0.p.s 0.r 0.q.t.t 0
10∗∗∗ p.q.r.s 0.p.s 0.r.q.t 0.t 10∗∗∗ p.q.r.s 0.p.s 0.r.q.t.t 0
n=2
2
n=4
22
n=6
33 2112 6∗
n=8
44 2222 3113 211112 (2, 2) (2, 2) 6∗ 2.2 6∗ 2 : .2 8∗
n = 10
55 2332 3223 4114 212212
221122 311113 21111112 (2, 2+) (2, 2+) (3, 2) (3, 2)
(2 1, 2) (2 1, 2) (2, 2) 1 1 (2, 2) 6∗ 3.3 6∗ 2 1.2 1 6∗ 3 : .3
6∗ 2 1 : .2 1 6∗ 2.2.2.2 6∗ 2.2 : 2.2 6∗ 2.2.2 0 : 2 6∗ 2.2 : 2 0.2 0
8∗ 2.2 8∗ 2 : .2 8∗ 2 0.2 0 8∗ 2 0 : .2 0 10∗
10∗∗ 10∗∗∗
n = 12
66 2442 3333 4224
5115 213312 222222 231132
312213 321123 411114 21122112
21211212 22111122 31111113 2111111112
(4, 2) (4, 2) (3, 3) (3, 3) (3, 2 1) (3, 2 1) (3, 2+) (3, 2+)
(3, 2) 1 1 (2, 3) (3 1, 2) (3 1, 2) (2 2, 2) (2 2, 2) (2, 2, 2) (2, 2, 2)
(2, 2 + +) (2, 2 + +) (2, 2) 2 2 (2, 2) ((2, 2), 2) ((2, 2), 2) (2, 2+) 1 1 (2, 2+)
(2, 2) 1 1 1 1 (2, 2) (2 1, 2 1) (2 1, 2 1) (2 1, 2+) (2 1, 2+) (2 1, 2) 1 1 (2, 2 1)
(2 1 1, 2) (2 1 1, 2) 6∗ 4.4 6∗ 3 1.3 1 6∗ 2 2.2 2
6∗ 2 1 1.2 1 1 ∗
6 (2, 2).(2, 2) 6∗ 4 : .4 6∗ 3 1 : .3 1
6∗ 2 2 : .2 2 6∗ 2 1 1 : .2 1 1 6∗ (2, 2) : .(2, 2) 6∗ 3.2.2.3
6∗ 2.3.3.2 6∗ 2 1.2.2.2 1 6∗ 2.2 1.2 1.2 6∗ 3.3 : 2.2
6∗ 3.2 : 3.2 6∗ 2 1.2 1 : 2.2 6∗ 2 1.2 : 2 1.2 6∗ 3.2.3 0 : 2
6∗ 2.3.2 0 : 3 6∗ 2 1.2.2 1 0 : 2 6∗ 2.2 1.2 0 : 2 1 6∗ 3.3 : 2 0.2 0
6∗ 2 1.2 1 : 2 0.2 0 6∗ 2.2.2.2.2.2 6∗ 2.2.2.2.2 0.2 0 6∗ 2.2.2 0.2.2.2 0
8∗ 3.3 8∗ 2 1.2 1 8∗ 3 : .3 8∗ 2 1 : .2 1
8∗ 3 0.3 0 8∗ 2 1 0.2 1 0 8∗ 3 0 : .3 0 8∗ 2 1 0 : .2 1 0
8∗ 2.2.2.2 8∗ 2.2 : 2 : .2 8∗ 2.2 : .2.2 8∗ 2.2.2 0 :: .2 0
8∗ 2.2 : 2 0 : .2 0 8∗ 2.2 : .2 0.2 0 8∗ 2.2 0.2 0.2 8∗ 2.2 0 : .2 0.2
8∗ 2 : 2 0.2 0 : 2 8∗ 2 0.2 0.2 0.2 0 8∗ 2 0.2 0 : 2 0 : .2 0 ∗
8 2 0.2 0 : .2 0.2 0
10∗ 2.2 10∗ 2 : .2 10∗ 2 ::: .2 10∗ 2 0.2 0
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
338 LinKnot
For future derivation, one may check achirality of all rational, stel-
lar, arborescent and polyhedral alternating source KLs using Kauffman
polynomials computed for a KL and its mirror image, the LinKnot func-
tions AmphiProjAltKL and AmphiAltKL and the program SnapPea
by J. Weeks, and then systematically derive classes of achiral KLs.
P.G. Tait conjectured that every achiral KL must have an even number
of crossings, so neither P.G. Tait nor M.G. Haseman considered the possibil-
ity of the existence of achiral knots with an odd crossing number. However,
this Tait’s Conjecture was disproved (see Hoste, Thistlethwaite, and Weeks,
1998). The first oriented achiral non-alternating link 8∗ . − 2 0.2 0. − 2 0 with
n = 11 crossings was discovered in 1998 (Liang, Mislow and Flapan, 1998).
The achiral non-alternating knot 10∗∗ 2 0.2 : −2 0 : 2 0.−1.−1.−1.−2 0 with
n = 15 crossings was found by M. Thistlethwaite, who also recognized few
repetitions in Haseman’s tables. However, Tait’s Conjecture about achiral
KLs holds for alternating KLs: there is no alternating achiral KL with an
odd number of crossings (Corollary 1.1).
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
340 LinKnot
with n = 16 crossings (Fig. 2.94), that shows its achirality. The non-
alternating achiral knot
10∗∗ 2 0.2 : −2 0 : 2 0. − 1. − 1. − 1. − 2 0
of the knot
10∗∗ 2 0 : 2 0. − 2. − 1.2 0 : 2 0. − 2. − 1
{{39}, {8, −32, −34, 64, −62, −60, −66, 42, 40, 38, −68, −70, 44, 46, 48, 50,
−4, −6, 24, 18, 16, 142, 6, 28, 30, 36, −72, −74, −76, −12, −10, −78}}.
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
There are infinitely many pairs of P -undetectable KLs with the same
or different number of crossings, and there is an infinite number of unde-
tectable KLs for any polynomial invariant P (see, e.g., Kanenobu, 1986;
Przytycki, 1995; Watson, 2004). L. Watson showed that an arbitrary tan-
gle T can be extended to produce diagrams of two distinct knots that can
not be distinguished by the Jones polynomial. For a prime tangle T , the
5 The function Kh computes Khovanov polynomial (Shumakovitch, 2004). In the same
sense, we will use the term “Links-Gould polynomial” (De Wit, 2000).
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
342 LinKnot
t2 are arbitrary fixed tangles, and x, y are integers (Fig. 2.99). This family
has the following properties:
(1) all KLs from F are Alexander- and Conway-undetectable for every x,
y with x = y (mod 2);
(2) all KLs from F are Jones-undetectable for every x, y satisfying the
condition |x − y| = const, for arbitrary constant const;
(3) the links L(x, y) and L(x′ , y ′ ) from F are Jones-, Khovanov-,
HOMFLYPT-, A2-, and Links-Gould-undetectable iff |x−y| = |x′ −y ′ |,
x = x′ (mod 2), and y = y ′ (mod 2).
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
344 LinKnot
346 LinKnot
Here is the list of families we have obtained, where x and y are para-
meters, and all other tangles are fixed (Figs. 2.99-2.100):
6 According to the Conway notation, the basic polyhedron 1449∗ from LinKnot data
348 LinKnot
Proposition 2.2. There exist knots with the same colored Jones polynomi-
als (for all colors), HOMFLYPT and Kauffman polynomials, volume and
signature, but different Khovanov (and reduced Khovanov) homology.
where the bar above the number of crossings means the mirror image of the
corresponding knot.
In order to generalize their results, we described ten classes of KLs,
given in Conway notation and enumerated by I-X, containing 32 mentioned
pairs of Khovanov-detectable almost mutant knots. Most of them are cabled
mutants, but few of them (3,9,10,21,24,29) are obtained by a more general
method: inserting tangles in the top crossings of the tangle T (1, n).
A non-alternating representation of KL is called non-reducible if it can
not be reduced to a KL with a lower number of crossings.
Conjecture 2.7. All non-reducible KLs from the classes I-X with arbi-
trary tangles a,b,..., x,y are almost mutant KLs with different Khovanov
polynomials.
In other words, the pair-classes are equivalence classes of almost mutant
KLs detectable by Khovanov homology.
Before describing particular classes, we will explain common construc-
tion principle: the tangle T = T (1, n) is rotated around vertical 2-axis,
giving the tangle T ′ . By composing T and T ′ with an arbitrary (n + 1)-
tangle T1 , we obtain two KLs (T, T1 ) and (T ′ , T1 ). If the tangle T1 consists
from 2-tangles a,b,..., and x and y are tangles placed in the top crossings
of T and T ′ , by appropriate choice of signs and positions of the constitut-
ing tangles a, b,...,x,y we obtain a class-pair of almost mutant KLs with
distinct Khovanov homology. All other “non-appropriate” choices giving
non-reducible KLs result in completely undetectable almost mutant KLs.
The pair-classes I-X are derived from four basic polyhedra, 112∗ , 127∗ ,
132∗ , and 137∗ .
350 LinKnot
a b c d e x y
I 1 2 1 1 1 (1)
I 1 21 1 1 1 (2)
VI 1 2 1 1 2 1 (3)
II 2 2 1 1 1 (4)
I 1 2 2 1 1 (5)
VI 1 21 1 1 1 1 (6)
I 1 22 1 1 1 (7)
I 21 2 1 1 1 (8)
I 1 21 1 2 1 (9)
II 2 2 1 2 1 (10)
II 2 21 1 1 1 (11)
VI 2 2 1 1 1 1 (12)
VI 1 2 1 2 1 1 (13)
VI 1 2 2 1 1 1 (14)
III 1 21 2 1 1 (15)
III 1 2 21 1 1 (16)
I 1 31 1 1 1 (17)
I 1 4 1 1 1 (18)
I 3 2 1 1 1 (19)
II 3 2 1 1 1 (20)
IV 2 1 2 2 1 (21)
VIII 1 2 1 2 1 1 (22)
IX 2 1 1 1 1 1 1 (23)
I 1 2 2 2 1 (24)
I 1 3 2 1 1 (25)
III 1 3 2 1 1 (26)
III 1 2 3 1 1 (27)
I 1 2 3 1 1 (28)
V 1 2 2 2 1 (29)
X 2 1 1 1 1 1 1 (30)
II 2 2 2 1 1 (31)
VII 1 2 2 1 1 1 (32)
The pair-classes are not disjoint. For example, the knot (1) can also be
obtained in the class II, or III for b = 2.
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
All classes I-X are constructed in the same way, as a closure of tangle
T (1, 2) by an antiprismatic k-belt consisting from k triangles (k = 1, 2, ...)
(Fig. 2.69). However, instead of antiprismatic belt, arbitrary (n + 1)-tangle
T1 can be used (see, e.g., the pair-class derived from the basic polyhedron
171312∗, Fig. 2.106). The same tangle T (1, n) and (n + 1)-tangle T1 used
as its closure can result in different basic polyhedra, according to different
pairing of free strands (e.g., the classes VI and VIII). Representing the pair-
classes I-X by KL diagrams with labelled tangles a,b,c,...,x,y, we obtain
five diagrams, corresponding to the class-pairs I-V,VI-VII,VIII,IX, and X,
respectively (Fig. 2.104).
Figure 2.105 shows the knot pair (9) with x = 2, that can not be derived
as a pair of cabled mutant knots, without introducing tangle x in the top
crossing of the generating tangle T (1, 2).
Following the same pattern, we can derive almost mutant KLs de-
tectable by Khovanov polynomial from other basic polyhedra with a higher
number of crossings. For example, we derived the following pair-classes:
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
352 LinKnot
Fig. 2.106 Diagram of the pair-class derived from the basic polyhedron 171312∗ .
{{25}, {6, 12, −22, 20, −30, 2, 46, −40, −34, 10, 50, −4,
−48, 42, 36, −8, 28, −18, 32, 26, −16, 38, −24, 14, −44}}
and
{{25}, {6, −12, 24, 22, −32, 20, −4, −48, 42, 36, 8, 2, 46, −40, −34,
−10, −38, 18, −30, −44, 16, −28, 50, −14, 26}}.
The open question is: are there some other constructions resulting in
Khovanov-detectable almost mutant KLs.
Polynomials are determined via LinKnot and the function Kh from
KnotTheory program. Hyperbolic volumes of knots are computed via
Knotscape, and those of links via SnapPea (possibly with an insufficient pre-
cision). For example, Khovanov-detectable almost mutant links obtained
in the class I for b = 3, a = c = x = y = 1 have equal multi-variable
Alexander polynomials and, conjecturally, the same volume (approximately
12.6684303...) (Fig. 2.108).
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
354 LinKnot
Fig. 2.107 (a) Diagram of the pair-class derived from the basic polyhedron 1912169∗ ;
(b) Khovanov-detectable almost mutant knots obtained from it for for b = 2, c = 2 and
a = d = x = y = u = v = 1.
Fig. 2.108 Khovanov-detectable almost mutant links obtained from pair-class I for b =
3, a = c = x = y = 1.
Fig. 2.109 Try to prove that the unknotting number of this knot, 16n
300395 , is 3.
{{19}, {−18, 24, −36, 14, 38, −28, 6, 8, −22, 30, −2, −16, −32, −34, −10, 20, −26, −4, 12}}
{{19}, {20, 24, −36, −12, 28, −38, −6, 22, 2, −30, 14, 16, −32, −34, 10, −18, −26, −4, −8}},
{{18}, {−6, −10, −24, −32, −2, −18, 20, 36, 34, −26, 28, 14, −4, −12, 22, 16, −8, 30}},
{{17}, {6, 10, −20, 14, 2, −30, 8, 28, 24, 32, −4, 16, 18, 34, −12, 22, 26}}.
356 LinKnot
J- Kh- H-
n=8 1 1
n = 10 2 2 1
n = 12 11 4 5
n = 14 30 20 10
n = 16 65 32 23
358 LinKnot
J- Kh- H- K-
n = 10 1 1 1 1
n = 12 2 2 1 1
n = 14 7 6 3 4
n = 16 21 17 4 5
The next class of KLs considered are non-alternating pretzel KLs. The
first chiral J- and H-undetectable 3-component link 4, 2, 2− with n = 8
crossings is followed by chiral P -undetectable KL with an odd number
of crossings, which are Kh-detectable. Chiral knot 2 2, 3, 2− (942 ) is K-
undetectable (the phenomenon called “ 942 syndrome” by C. Liang and
K. Mislow, 1995), but Kh-detectable. For 8 ≤ n ≤ 16 we have the following
results:
J- Kh- H- K-
n=8 1 1
n=9 2 1 1
n = 10 3 1
n = 11 3 1 1
n = 12 14 3 6
n = 13 6 3 4
n = 14 40 7 11
n = 15 14 6 5
n = 16 83 22 27
360 LinKnot
Fig. 2.111 (a) Achiral links −3, −4, −5, 3, 4, 5 with a centro-antisymmetric diagram,
and −3, 3, −4, 5, −5, 4 with a mirror-antisymmetric diagram; (b) diagram of chiral link
−3, −5, 3, 4, 5, −4.
3-component link (2, 4) (3, 3), for n = 16 J-, H-, and Kh-undetectable knots
(2 1 2 1 1, 2) (3, 2), (2, 5) (3, 4), and J- and Kh-undetectable 2-component
links (2, 3 1) (3, 4 1), (3 1 2, 2) (4, 2). All of them are K-detectable.
In the class of K-, H-, and Kh-undetectable alternating knot
(2, 2 1+) (3, 3 1+) with n = 14 crossings, for every even n ≥ 16 we have
chiral P -undetectable KLs, which are undetectable by any P -polynomial.
The situation is similar with non-alternating algebraic KLs. For ex-
ample, for n = 12 there is non-alternating J-, H- and Kh- undetectable
3-component link (2, 2) (6, 2−), for n = 13 J- and Kh-undetectable 2-
component link (2, 2 1) (2 3, 3−), for n = 14 the same class contains 12 J-,
6 H-, and 5 Kh-undetectable chiral KLs. Those five KLs can be detected
as chiral only by K-polynomial.
All P -polynomials fail in the case of chiral knot (2 1, 2+) (3, 2−) with
n = 11 crossings, etc.
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
Table 12
n = 10 2 2, 2 1, 2+
n = 12 2 1 1 1, 2 2, 2+
n = 14 2 1 1 1 1 1, 2 2, 2+ 2 1 1 1 2, 2 1, 2 1+ 2 1 1 2, 2 1 1, 2 1+
2 2 2 2, 2 1, 2+
n = 16 2 1 1 1 1 1 1 1, 2 2, 2+ 2 1 1 2, 2 1 1 1, 2 1 1+ 2 2 1 2, 2 1 2 1, 2+
2 2 2 1 1, 2 2, 2 1+ 2 2 2 2, 2 1 1 1, 2+
n = 18 2 1 1 1 1 1 1 1 1 1, 2 2, 2+ 2 1 1 1 1 1, 2 1 1 2, 2 1 1+ 2 1 1 1 1 1 2, 2 1 1 1, 2 1+
2 1 1 1 1 2, 2 1 1 1 1, 2 1+ 2 2 2 1 1, 2 1 1 1, 2 2+ 2 2 2 2, 2 1 1 1 1 1, 2+
2 2 2 2 2 2, 2 1, 2+
n = 20 2 1 1 1 1 1 1 1 1 1 1 1, 2 2, 2+ 2 1 1 1 1 1 1 1, 2 1 1 2, 2 1 1+ 2 1 1 1 1 1 1 1, 2 2 2 2, 2+
2 1 1 1 1 2, 2 1 1 1 1, 2 1 1 1+ 2 1 1 2 1 1, 2 2 1 1 1, 2 2+ 2 1 1 2 2 1 1 2, 2 1 1, 2 1+
2 2 2 1 1, 2 1 1 1 1 1, 2 2+ 2 2 2 2 2 2, 2 1 1 1, 2+
n = 16 (2 1, 2+) (2 1 1 1 1 1, 2+)
n = 18 (2 1 1 1 1 1 1 1, 2+) (2 1, 2+) (2 1 1 1 1 1, 2+) (2 1 1 1, 2+)
(2 1 1 1, 2 1 1+) (2 1 1, 2 1+)
n = 20 (2 1 1 1 1 1 1 1 1 1, 2+) (2 1, 2+) (2 1 1 1 1 1 1 1, 2+) (2 1 1 1, 2+)
(2 1 1 1 1 1, 2 1 1+) (2 1 1, 2 1+)
n = 22 (2 1 1 1 1 1 1 1 1 1 1 1, 2+) (2 1, 2+) (2 1 1 1 1 1 1 1 1 1, 2+) (2 1 1 1, 2+)
(2 1 1 1 1 1 1 1, 2 1 1+) (2 1 1, 2 1+) (2 1 1 1 1 1 1 1, 2+) (2 1 1 1 1 1, 2+)
(2 1 1 1 1 1, 2 1 1+) (2 1 1 1, 2 1 1+) (2 1 1 1 1, 2 1 1 1+) (2 1 1 1 1, 2 1+)
n = 24 (2 1 1 1 1 1 1 1 1 1 1 1 1 1, 2+) (2 1, 2+) (2 1 1 1 1 1 1 1 1 1 1 1, 2+) (2 1 1 1, 2+)
(2 1 1 1 1 1 1 1 1 1, 2 1 1+) (2 1 1, 2 1+) (2 1 1 1 1 1 1 1 1 1, 2+) (2 1 1 1 1 1, 2+)
(2 1 1 1 1 1 1 1, 2 1 1+) (2 1 1 1, 2 1 1+) (2 1 1 1 1 1, 2 1 1 1 1+) (2 1 1 1 1, 2 1+)
362 LinKnot
n=9 2 2, 3, 2−
n = 11 2 1 2, 2 2, 2−
n = 13 2 1 1 1 2, 2 2, 2− 2 1 1 1 1 1, 3, 3− 2 1 1 2, 2 1 1, 3−
2 2 2 2, 3, 2−
n = 15 2 1 1 1 1 1 2, 2 2, 2− 2 1 1 2, 2 1 2, 2 1 1− 2 1 2 2, 3 1 2, 2−
2 2 2 1 1, 2 2, 3− 2 2 2 2, 2 1 2, 2−
n = 17 2 1 1 1 1 1 1 1 2, 2 2, 2− 2 1 1 1 2, 2 1 1 2, 2 1 1− 2 1 1 1 1 1 2, 2 1 2, 3−
2 1 1 1 1 2, 2 1 1 1 1, 3− 2 2 2 1 1, 2 1 2, 2 2− 2 2 2 2, 2 1 1 1 2, 2−
2 2 2 2 2 2, 3, 2−
n = 19 2 1 1 1 1 1 1 1 1 1 2, 2 2, 2− 2 1 1 1 1 1 2, 2 1 1 2, 2 1 1− 2 1 1 1 1 1 2, 2 2 2 2, 2−
2 1 1 1 1 2, 2 1 1 1 1, 2 1 2− 2 2 1 1 2, 2 1 2 1 1, 2 2− 2 1 1 2 2 1 1 2, 2 1 1, 3−
2 2 2 1 1, 2 1 1 1 2, 2 2− 2 2 2 2 2 2, 2 1 2, 2−
Fig. 2.112 (a) Chiral knot 10∗∗∗ 2 :: .2 0 : .2 0.2; (b) achiral knot 10∗∗∗ .2 0 :: 2 0 : .2 0.2.
We hope that future KL tables will follow the vertical structure (fami-
lies of KLs), and not the minimal crossing number of KLs (“horizon-
tal” structure). The concept of new KL tables, given in Appendix A
(http://www.mi.sanu.ac.yu/vismath/Appendix.pdf), is now still restricted
to generators of KL families with n ≤ 9 crossings, but we hope that very
soon will be extended to generating KLs with a larger number of crossings.
364 LinKnot
and the only tables containing links with n ≤ 9 crossings are given by
D. Rolfsen (1976). Tables can be found in the books by D. Rolfsen (1976),
G. Burde and H. Zieschang (1985), L.A. Kauffman (1987a), C. Adams
(1994), A. Kawauchi (1996), K. Murasugi (1996), and V. Manturov (2004).
366 LinKnot
Knot family: 2k + 1
Notation:
3 31
5 51
7 71
9 91
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
368 LinKnot
Dowker codes:
3 4 6 2
5 6 8 10 2 4
7 8 10 12 14 2 4 6
9 10 12 14 16 18 2 4 6 8
11 12 14 16 18 20 22 2 4 6 8 10
13 14 16 18 20 22 24 26 2 4 6 8 10 12
15 16 18 20 22 24 26 28 30 2 4 6 8 10 12 14
17 18 20 22 24 26 28 30 32 34 2 4 6 8 10 12 14 16
19 20 22 24 26 28 30 32 34 36 38 2 4 6 8 10 12 14 16 18
Alexander polynomials:
3 [1 −1
5 [1 −1 1
7 [1 −1 1 −1
9 [1 −1 1 −1 1
11 [1 −1 1 −1 1 −1
13 [1 −1 1 −1 1 −1 1
15 [1 −1 1 −1 1 −1 1 −1
17 [1 −1 1 −1 1 −1 1 −1 1
19 [1 −1 1 −1 1 −1 1 −1 1 −1
2k
X
∆(p) = (−1)i ti
i=0
Jones polynomials:
3 1 4 1 0 1 −1
5 2 7 1 0 1 −1 1 −1
7 3 10 1 0 1 −1 1 −1 1 −1
9 4 13 1 0 1 −1 1 −1 1 −1 1 −1
11 5 16 1 0 1 −1 1 −1 1 −1 1 −1 1 −1
13 6 19 1 0 1 −1 1 −1 1 −1 1 −1 1 −1 1 −1
15 7 22 1 0 1 −1 1 −1 1 −1 1 −1 1 −1 1 −1 1 −1
17 8 25 1 0 1 −1 1 −1 1 −1 1 −1 1 −1 1 −1 1 −1 1 −1
19 9 28 1 0 1 −1 1 −1 1 −1 1 −1 1 −1 1 −1 1 −1 1 −1 1 −1
Symmetry group: D1
Symmetry type: chiral, invertible.
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
Signature: 2k
Unknotting number: k
The concept of new knot tables, based on knot families, can be naturally
extended to links. Working with the program LinKnot we succeeded to
extend new knot tables to all KLs with n ≤ 9 crossings. Tables given in the
Appendix A (http://www.mi.sanu.ac.yu/vismath/Appendix.pdf) contain
generating KLs for every family of KLs, the Conway symbol of the family
with conditions for parameters, the number of components, the Alexander
polynomial given by general formula, general formulas for the number of
different projections (only for alternating KLs), general formulas for BJ-
unknotting (BJ-unlinking) numbers and signatures, data about period(s)
of KLs, achirality, and BJ-unlinking gap.
The concept of new KL tables is based on the notion of generating KLs
and families originating from them. Hence, one of the possible future goals
is a search for new KL invariants that will be able to recognize families.
New “family” invariants have to be preserved by Reidemeister moves and n-
moves, since all members of some family can be obtained from a generating
KL by a sequence of n- moves. Unfortunately, the transition from one
family member to another is only possible at the level of minimal canonical
Conway symbols. In this way we have obtained KL tables given in the
Appendix A, that consist only of generating KLs, families derived from
them, and parametric data about families (i.e., the KL properties and
invariants in a general form). Computational results imply that Alexander
polynomials of two alternating KLs from the same family must be different–
therefore all we need for distinguishing alternating KLs would be their
family and Alexander polynomial. The solution is even simpler if we restrict
our attention to alternating KLs given in Conway notation– we can use a
minimal Dowker code (obtained directly from the Conway symbols by the
LinKnot function MinDowAltKL). This function gives minimal Dowker
code for all alternating KLs, except for those derived from basic polyhedra
permitting flypes. If a projection of a basic polyhedron permits flypes, we
are not able to compute the minimal Dowker codes for all alternating KLs
derived from this basic polyhedron.
370 LinKnot
Fig. 2.115 Chain of bigons (a) after crossing change (b); (c) antiparallel case; (d) parallel
case.
372 LinKnot
Chapter 3
The discovery of knots probably predates that of fire or wheel. Ropes, cords,
and knots needed to secure them played an important role in the early tech-
nological development. The main reason for the lack of discovery of such
artifacts is that they have been made from organic materials (vegetable
fibres, sinews, thongs, hair, etc.), thus subject to decay. However, even cer-
tain wild gorillas are able to make complete knots, primarily Granny and
Reef knots, so that the beginning of knot tying most likely preceded the
evolution of mankind. The indirect testimony for an early use of cordage
and knots are perforated objects, beads or pendants, dating some 300 000
years ago, and spherical stones found in Africa and China (about 500 000
years old), probably used as bola weights in hunting. More recently bows
and arrows that required well-made cordage and secure knots, as well as Pa-
leolithic figurine in soft limestone from Kostenki (Russia, 24 000 B.C.) show
belts made from multiple twined flexible elements. Some actual Neolithic
knots are preserved in North Zealand and Denmark. Sophisticated plaits
made with strips of date palm leaf originate from Ancient Egypt (Turner
and Van De Griend, 1995). Arrangements of knots served as a basis for
mathematical recording systems in the Peruvian quipus or Zuñi knots from
the New Mexico, where the knots functioned as symbolic and mnemonic
devices. Various examples of knot-art can be found in all ancient civiliza-
tions, in Japanese and Chinese art, Celtic art, ethnic Tamil and Tchokwe
art, in Arabian, Greek or Smyrnian laces... Celts made extensive use of
knot-work pictures created for decorative and religious purposes (G. Bain,
1973; I. Bain, 1990). Their art required a high level of mathematics, to
geometrically create knotted curves even with zoomorphic ornaments.
375
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
376 LinKnot
1 The first application of geometria situs dates from 1736: it is L. Euler’s solution of the
famous Köningsberg bridges problem, that represents the beginning of graph theory.
2 H.K. Brunn observed in 1892 that the linking number of two-component link, consid-
ered by Gauss, can be read from a diagram of the link (Definition 1.32).
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
The main inspiration for enumerations of knots come from Sir William
Thomson (Baron Kelvin of Largs). In attempt to make a classification of
chemical elements, in the mid-1860s he announced his model of the atom,
called “vortex theory”. After Helmholtz’s paper (1858) on vortex motion,
based on an experiment with vortex smoke rings, Kelvin developed the the-
ory of particles as tiny topological twists, i.e., knots in the ether (1867). He
believed that the variety of chemical elements can be explained by the kinds
of different knots. He also thought that the ability of atoms to transform
into each other, transmutation, was related to the cutting and recombin-
ing of knots. This theory was taken seriously until the wide acceptance of
Mendeleev’s periodic system of elements by the scientific community.
Vortex theory inspired P.G. Tait to start with the enumeration and clas-
sification of knotted structures and solve the census problem. For this, he
developed Scheme-method, a representation of reduced knot diagrams by
codes (already known to Gauss) and the Partition method (Turner and Van
De Griend, 1995), an improvement of Listing’s attempt. In collaboration
with the Reverend T.P. Kirkman and C.N. Little, they succeeded in making
a list of all alternating knots up to 11 crossings. The derivation of knots with
10 crossings took them six years to complete. Tait also considered some of
the fundamental problems in knot theory: chirality (Definition 1.26) and
unknotting number (or Gordian number, called by Tait “beknottedness”)
(Definition 1.56), and introduced the graph of a knot (page 24). He made
a few conjectures on alternating knots, e.g., that the minimal number of
crossings of an alternating KL is always realized in an alternating diagram
(Kauffman-Murasugi Theorem, Theorem 1.6), and that two minimal dia-
grams of the same oriented alternating KL have the same writhe (Theorem
1.7). His famous Flyping Conjecture was recently proved by Menasco and
Thistlethwaite (1991, 1993), about 100 years after it was formulated (The-
orem 1.11). Kirkman’s geometrical system for the systematic derivation of
knot projections (4-valent planar graphs) was closely related to the enumer-
ation of basic polyhedra and, at the same time, represented a geometrical
method for classifying knot projections. Kirkman derived the census of
1581 plane curves with 11 crossings from which Little distinguished 357
different alternating knots. Little also considered the derivation of non-
alternating knots, and in addition to the flype (page 44), introduced a
2-pass: a KL transformation where a string is simply pulled over a tangle
(page 47). After six years of work, Little produced a catalogue consisting
of 43 non-alternating knots with n = 10 crossings and 551 drawings of their
various minimal projections (with few omissions). The only serious error in
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
378 LinKnot
his tables was the duplication discovered by K. Perko in 1974 (Fig. 1.26).
Little observed that the writhe of a reduced knot diagram is invariant with
respect to flypes and 2-passes, and proposed that it is a knot invariant, but
it is not: the first known counterexample is Perko pair. Little erroneously
believed that just two kinds of moves, flypes and 2-passes, are sufficient to
pass between all minimal diagrams of the same knot3 . After Tait, Kirk-
man and Little, pertaining to knot tabulation, almost nothing important
happened for a century, until the works of J.H. Conway and A. Caudron in
1970-80s, and the computer derivation of KLs.
The principal problem in knot enumeration is deciding when two knots
are ambient isotopic (Definition 1.20). Two KLs are isotopy equivalent if
one of them can be transformed to the other by pushing and pulling, but not
cutting, its string(s). The problem of isotopy, known as the Knot problem,
became the main problem in knot theory.
Closely connected to the knot problem is the problem of achirality, am-
bient isotopy of a KL to its mirror image (Definition 1.26). Thirty years
after Tait’s first results in enumeration of achiral knots with n ≤ 10 cross-
ings, M.G. Haseman in her dissertation partially extended knot tables, and
described achiral knots with n ≤ 12 crossings. Tait conjectured that every
achiral KL must have an even number of crossings. Therefore neither Tait
nor Haseman considered the possibility of the existence of achiral knots with
an odd crossing number. The first oriented achiral link 8∗ .−2 0.2 0.−2 0 with
n = 11 crossings was discovered in 1998 (Liang, Mislow and Flapan, 1998).
The achiral non-alternating knot 10∗∗ 2 0.2 : −2 0 : 2 0.−1.−1.−1.−2 0 with
n = 15 crossings was found by M. Thistlethwaite, who also recognized sev-
eral duplicates in Haseman’s tables. However, the Tait’s Conjecture about
achiral KLs holds for alternating KLs: there is no alternating achiral KL
with an odd number of crossings (Corollary 1.1).
After the empirical phase, the emphasis in the theory of knots turned
away from enumeration toward attempts to prove the completeness of knot
lists and to show that they do not contain repetitions. The first steps in
the development of the required mathematical apparatus were made by
H. Poincaré, who introduced several topological objects and tools, e.g.,
the concept of the complex and its fundamental group. The first proof of
the existence of non-trivial knots is given by H. Tietze in 1908, using the
fundamental group of knot complement (or knot group). W. Wirtinger in
3 In the derivation of non-alternating knots and in the knot minimization program
knotfind.c (the part of Knotscape) M. Thistlethwaite used 13 different diagrammatic
moves (Hoste, Thistlethwaite and Weeks, 1998).
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
380 LinKnot
382 LinKnot
tion functions. Hence, Jones did not only introduced a refined invariant for
distinguishing and analyzing KLs, but related knot theory to other fields
of mathematics and theoretical physics, in particular statistical mechanics
and quantum field theory. This revolutionary discovery was followed by
new more powerful polynomials: HOMFLYPT and Kauffman polynomial.
However, none of these polynomials is a complete invariant, able to distin-
guish all KLs, or to completely recognize chirality. Khovanov homology
(Khovanov, 1997, 2001), which categorifies one-variable Jones polynomial
is a strictly stronger invariant then the Jones polynomial itself (Bar Natan,
2002; Manturov, 2004).
The introduction of Vassiliev’s invariants gave rise to the hope that a
complete KL invariant can be found. Working in the space of knots, Vas-
siliev invariants are essentially different from all other previously mentioned
KL invariants: instead of associating to each KL an mathematical quan-
tity or polynomial, they assign to a KL a numerical value depending on a
set of initial conditions. Many of the invariants introduced before, such as
Alexander, Jones, and Kauffman polynomials, are Vassiliev invariants. On
the other hand, none of the classical KL invariants: the minimal crossing
number, unknotting (unlinking) number, signature, bridge number, braid
index and genus of a KL, are Vassiliev invariants.
The initial interest in knot theory was stimulated by Kelvin’s theory of
atomic structure (1867). By the turn of the century, after scientific confir-
mation of Mendeleev’s periodic tables, it was clear that Kelvin’s theory was
incorrect. Chemists were no longer interested in classifying knots. However,
topologists continued to study knots.
The focus of chemists turned towards attempts to synthesize molecular
KLs. The first pair of linked rings in a form of the Hopf link, a catenane,
was synthesized by H. Frisch and E. Wasserman in 1961. The first mole-
cular knot, a trefoil made out of 124 atoms was produced by C. Dietrich-
Bushecker and J.-P. Sauvage in 1989. They refer to stereochemical topology,
synthesis, characterization, and analysis of topologically interesting molec-
ular structures (Flapan, 2000).
Construction of numerous KLs become possible after the synthesis of
first molecular Möbius ladder with three rungs by D. Walba, R. Richards
and R.C. Haltiwanger in 1982, and addition of twists to the Möbius ladders
managed by Q.Y. Zheng in 1990 (Fig. 3.1). In fact, after breaking the
rungs, Möbius multi-strand twisted ladders became a molecular closed braid
representation of a KL.
In the 1950s F.H.C. Crick and J.D. Watson unravelled the double helix
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
structure of DNA. A molecule of DNA can also take the form of a ring and
become knotted. In the process of recombination, a DNA knot is temporar-
ily broken, physically changed, and then reconnected. In the 1970s it was
discovered that an enzyme, topoisomerase, is responsible for this process.
The first electron microscope pictures of knotted DNA were produced in
1985 (Wasserman, Dungan and Cozzarelli, 1985). The linking number and
its splitting into average writhe W r and twist T w is used as a basic tool
to analyze the geometry of supercoiled DNA. C. Ernst and D.W. Sumners
(1990, 1999) reconstructed the actions of enzyme (TN3 Resolvase) by solv-
ing tangle equations. Distances of rational knots and links were calculated
by I.K. Darcy and D.W. Sumners (2000).
Mathematical models or descriptions of large particle systems, such as
the Ising and Potts model, or Yang-Baxter equation, give rise to knot in-
variants generated by partition function.
Different applications of knot theory in physics, chemistry, and biology
are considered in books by L. Kauffman (1991), C. Adams (1994), K. Mura-
sugi (1996), E. Flapan (2000), and in the collection of papers edited by
D.W. Sumners (1993). In this book we will try to emphasize the beauty,
universality and diversity of knot theory through its various, non-standard
applications to ornamental art, fullerenes, self-referential systems, and KL
automata.
384 LinKnot
386 LinKnot
388 LinKnot
curve– single line placed uniformly in a regular tiling. How can we arrange
generating mirror sets and classify curves obtained? In principle, any poly-
omino (polyiamond or polyhexe) (Golomb, 1994) with mirrors on its bor-
der, and two-sided mirrors between cells or perpendicular to the internal
cell-edges in their midpoints, can be used for creating perfect curves.
We propose the following construction from a polyomino (polyiamond
or polyhexe):
(1) any mirror placed in a crossing point of two distinct curves connects
them in one curve;
(2) depending on the position of a mirror, a mirror placed into a self-
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
390 LinKnot
crossing point of an (oriented) curve either does not change the number
of curves, or breaks the curve in two closed curves (Fig. 3.7).
Fig. 3.7 A mirror placed in a crossing point of (a) two different curves; (b) one curve.
Fig. 3.8 The successive introduction of internal mirrors in the RG[2, 2] that preserves
a single curve.
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
Definition 3.1. Two mirror curves are equal iff there is a similarity trans-
forming one into the other.
In other words, one mirror curve can be obtained from the other by
a combined action of proportionality and isometry. Instead of considering
equality of curves, we may consider the equality of mirror arrangements
defined in the same way. In Subsection 3.2.4 we will try to find the number
of different perfect curves (i.e., their corresponding mirror arrangements)
which can be derived from RG[a, b] for a given number of mirrors m (m =
k − 1, k, . . . , 2ab − a − b).
392 LinKnot
Fig. 3.9 Celtic friezes with the same symmetry group of bands p1a1.
Fig. 3.10 Celtic friezes with the same symmetry group of bands p121.
and 80 symmetry groups of layers (see, e.g., Shubnikov and Koptsik, 1974;
Coxeter and Moser, 1980; Grünbaum and Shephard, 1986; Martin, 1980;
Jablan, 2002).
These criteria are not always sufficient, and we also need to consider other
knot-theoretical properties such as whether a pattern represents prime or
composite KL arrangement.
In the case of an n-tangle (n = 2, 3, ...) representing a fundamental
region, it is possible to construct all its closures and use them for a further
classification. For example, by joining left and right ends of the tangles (Fig.
3.10) that are the basic elements of two friezes with the same symmetry
p121 we obtain links 3#2 and 2 1 2, 2, 2+, respectively.
This classification, proposed by S. Jablan and Lj. Radović in 2001, is
similar to the approach proposed earlier by I. Emery (1995).
394 LinKnot
(I) If only k − 1 edge-incident mirrors are placed in RG[a, b], such that
a 6= b, the number of perfect curves is
k−3 k−1
(1) (4k)k−2 tk−1 + 2(4k) 2 t 2 for k odd;
k−2 k−2
(2) (4k)k−2 tk−1 + (4k) 2 zt 2 , with z = x for a ≡ 0 (mod 2k), b ≡ k
(mod 2k), and z = x + y, for a ≡ b ≡ k (mod 2k), for k even.
Even for relatively small RGs (e.g., with a = 6, b = 3), and minimal
number of mirrors (k−1 = 2), the number of the different curves obtained is
large. For example, there are 52 different arrangements of two edge-incident
mirrors in a rectangle 6 × 3 producing perfect curves. Among them, only 8
are symmetric– 4 mirror-symmetric and 4 point-symmetric (Fig. 3.11).
3.2.6 Polyominoes
A plane region without “holes”, formed by n edge-to-edge adjacent squares
is called a polyomino (Golomb, 1994). If instead of squares we use n equilat-
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
396 LinKnot
Fig. 3.12 (a) Possible border situations; (b) possible situations between vertical and
horizontal neighboring grid points; (c) Lunda designs.
398 LinKnot
These rules can be efficiently used for the computer enumeration of poly-
ominoes. In each step we need to derive (n + 1)-minoes from n-minoes by
adding a square, then check the equality of obtained polyominoes and make
the list of all (n + 1)-minoes. The main problem are “undesired” edge con-
tacts (e.g., contacts in parallel edges, producing “hollow” and “overlapping”
polyominoes).
400 LinKnot
Fig. 3.14 Hopf link 211 (2) and trefoil knot 31 (3) in RG[2, 2].
Fig. 3.15 Construction of a single mirror curve from the tiling (a) by connecting edge
mid-points (b), tracing components (c) and introducing a mirror (d).
402 LinKnot
Fig. 3.16 Lunda designs on the octahedron. Small triangles in uppermost right corners
show the coloring of an outside region.
tained iff a, b are mutually prime numbers. From the knot theory point
of view, every single-curve plate, turned into an alternating knot by in-
troducing the relation “over-under”, represents a Lissajous knot (Bogle,
Hearst, Jones and Stoilov, 1994). The infinite series of plates, obtained
for an arbitrary a (a ≥ 3) and b = 2, consists of the rational KLs of the
form 3 1 3, 3 1 2 1 3, 3 1 2 1 2 1 3, 3 1 2 . . . 2 1 3 (Fig. 3.17). Notice that for
every odd b we obtain a knot, and for every even b a 2-component link.
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
The number of different projections of these KLs is: 1, 4, 13, 68, 346,. . .,
respectively, but in knotworks, only one of them– the most symmetric, is
used for each a. Most symmetric projections can be found and drawn us-
ing the LinKnot function MaxSymmProjAltKL. For a = 3 we have the
projection 3 1 3, for a = 4 the projection (((1, (3, 1), 1), 1), 1, 1, 1), for a = 5
the projection (((1, ((1, (3, 1), 1), 1), 1), 1), 1, 1, 1), for a = 6 the projection
(((1, ((1, ((1, (3, 1), 1), 1), 1), 1), 1), 1), 1, 1, 1) etc. The sequence 1, 4, 13, 68,
346,. . . is not included in the Encyclopedia of Integer Sequences; in fact, it
is possible to obtain many new infinite sequences defined by numbers of
different projections of specific classes of KLs. For an arbitrary a (a ≥ 3)
and b = 3 we obtain plates with polyhedral KLs: for a = 3 we have 3-
component link 8∗ 2 : 2 : 2 : 2, for a = 4 the knot 1312∗ : 2 0 ::: 2 0.2.2 0, etc.
(Fig. 3.18).
Let us now describe four general rules for combining plate designs
and/or mirror curves. The first three rules are given by P. Gerdes (1999),
and the fourth is proposed by S. Jablan. We will restrict our consideration
to mirror curves placed in polyominoes with square cells.
Construction rules:
(1) The first rule defines a combination of two mirror curves that share
one edge of an open cell on their borders (Fig. 3.19a). Such a composition
corresponds to the direct product of KLs, and it was probably one of the
most exploited constructions in knotwork art. For given mirror curves M1
and M2 , this kind of direct product we will call ×-direct product and denote
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
404 LinKnot
Fig. 3.19 Rules for composing plate designs and mirror curves.
(3) The third rule is restricted to plate designs: two monolinear plate
designs whose overlapping contains exactly two cells will give a new mono-
linear plate design. The schematic interpretation of the third rule is given
in Fig. 3.19c.
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
406 LinKnot
408 LinKnot
The first rule, ×-direct product, is very frequently used in Celtic knot-
work art for the construction of friezes, as well as the k-direct product
(Rule 2). Both of them are the standard tools for obtaining translational
repetitive structures: friezes or even plane symmetry groups.
Applying the second rule actually connects two monolinear RGs in their
corners, i.e., it is the k-direct product of the corresponding knots. Another
possibility is using (more or less) “open” RGs and their k-direct product.
Although we obtain the same composite KLs, in the visual sense obtained
patters will be different. The ×-direct product was used in Celtic knot art
as well, mainly for the construction of frieze knotworks (or bordures).
In order to analyze Celtic knotworks based on k-direct product first we
need to insert some internal mirrors perpendicular to the edges in basic
(monolinear) RGs, in order to obtain parts or “tangles” of Celtic knot-
works with an appropriate placement of incoming and outgoing strands.
The possible choices for their positions are two top (or bottom) corners of
an elementary RG, two diagonal (ascending or descending) corners, or all
four corners forming a tangle. In the first case we place internal mirrors
perpendicular to “vertical” and “horizontal” edges of border cells, form-
ing an L-shape form (Fig. 3.21). Furthermore, cutting the long edge(s) of
the design and reconnecting them, we obtain different frieze designs (direct
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
410 LinKnot
Fig. 3.28 (a) k-direct product in the Tchokwe design; (b,c) the application of Rule 3 in
Tchokwe designs.
Fig. 3.29 Derivation of Celtic monolinear cross knot design from plate design obtained
using the Rule 4.
The third rule was one of the favorite rules in the construction of
Tchokwe sand drawings. A whole series of “social” monolinear plate de-
signs representing a leopard with cubs (Fig. 3.21a), a design called kambava
wamulivwe that represents an animal called kambava that died inside a rock
(Fig. 3.28b), or lusona drawing called tambwe that represents a lion (Fig.
3.28c) is composed in this way.
The fourth rule offers the highest degree of freedom and often gives
symmetric plates in knotwork art (Fig. 3.29). Various designs can be
obtained by adding along edge b any RG(a, b) with the property b|a, or
square RG, in a symmetric or asymmetric way to a monolinear plate design.
In this way, we can create perfect curves of a desired shape.
Creating a variety of monolinear plate designs opens the door to artis-
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
412 LinKnot
Fig. 3.33 Construction of Celtic monolinear knot design (b) by breaking the symmetry
of the two-component symmetric design (a).
tic creativity and play: there is a huge number of ways for introducing
internal edge-incident and edge-perpendicular mirrors in order to preserve
monolinearity (Figs. 3.30-3.33).
Together with the remarkable example of a monolinear cross knot design
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
Fig. 3.34 (a) Tchokwe sand drawings; (b) the geometric construction of the correspond-
ing basic polyhedra.
414 LinKnot
Fig. 3.38 (a) A tangle T , its numerator N (T ), and denominator closure D(T ); (b) the
pair of links (2 1 2, 2 1 2) obtained as the numerator and denominator closure of the Celtic
tangle; (c) the closure of the tangle 3; (d) the bands (p121,3) and (p121,5).
In the same way we can obtain different basic polyhedra derived from
shadows of torus knots [a,b], GCD(a, b) = 1. They represent the same
geometric structure: a series of inscribed n-gons (n ≥ 3).
Similar infinite series of basic polyhedra inspired by patterns from na-
ture, such as the growth patterns of certain plants, can be found in art-
works. For example, a shadow of the torus knot [12,5] (Fig. 3.35) appears
in Michelangelo’s plaza (Fig. 3.36).
Celtic masters used friezes without bigons to construct basic polyhedra
(Fig. 3.37b) by identifying opposite sides of friezes. This method has quite
a general character and can be used for creating other circular knot designs
(Fig. 3.27, 3.37a).
In order to classify complex periodic knotworks (e.g., Celtic friezes or
plane ornaments from Fig. 3.26, or laces), we will recognize basic patterns–
tangles, equivalent to fundamental regions and combine two approaches:
the theory of symmetry and knot theory. First we determine the symmetry
group, and then add the information about tangles. For this description,
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
416 LinKnot
418 LinKnot
Fig. 3.41 (a) Two self-avoiding curves derived from Borromean rings; (b) self-avoiding
curve derived from the fullerene C60 by the mid-edge truncation.
Fig. 3.42 (a) Coding of the first self-avoiding curve from Fig. 3.41a and its chord
diagram; (b) chord diagram of the second self-avoiding curve from Fig. 3.41a and its
dual.
3.42a, or by its dual obtained by inverse bicoloring, where full (black) lines
are replaced by broken (white) lines and vice versa. In the both cases (Fig.
3.42) chord diagrams are equal to their duals, i.e., they are self-dual. From
every chord diagram we can easily obtain a code of the corresponding self-
avoiding curve and vice versa, and to recover an original KL from which
the curve is derived. Two stages of the recovering are illustrated in Fig.
3.43.
420 LinKnot
n n
2n k
X (−1)k k [ 2n−k ]2 (2n − k)!
.
2n n!
k=0
Chord diagrams derived for n = 2, 3, 4 are given in Fig. 3.44. Among all
chord diagrams we can distinguish 2-vertex connected graphs (containing
the edges of an 2n-gon as well), corresponding to non-prime KLs, and
others, 3-connected, corresponding to prime KLs.
For coloring of chord diagrams we have the rule: every two diagonals
crossing each other must have different colors. A chord diagram will be
colorable iff it is planar. The other, purely visual, criterion for colorability
is the following: a chord diagram is colorable iff crossings of its diagonals
do not form a polygon with an odd number of edges, and three or more
diagonals do not have a common point (Fig. 3.45). Coloring of a (colorable)
chord diagram represents a projection of a polyhedron enclosed in an 2n-
gon, with proper visibility of all edges.
In the case of 2-vertex connected chord diagrams, coloring is not unique:
from the same uncolored chord diagram we can obtain several different col-
ored diagrams (Fig. 3.44). KL shadows, their corresponding self-avoiding
curves and colored chord diagrams for n = 2, 3, 4 are given in Fig. 3.46,
and for n = 5 in Fig. 3.47. In the case of 3-vertex connected chord di-
agrams, a coloring is completely forced by the coloring of one edge: by
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
choosing its color we can obtain only one colored chord diagram, or its
dual. Hence, in the case of 3-vertex connected planar diagrams, an un-
colored chord diagram provides a complete information about the cor-
responding self-avoiding curve. Every uncolored chord diagram can be
given as a list of unordered pairs of numbers denoting chords. For ex-
ample, the uncolored chord diagram from the Fig. 3.48a can be denoted as
{{1, 3}, {2, 6}, {4, 9}, {5, 8}, {7, 10}}. The same figure illustrates its bicol-
oring (a), the reconstruction of its corresponding self-avoiding curve (b-e),
and KL shadow obtained (f).
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
422 LinKnot
Fig. 3.46 KL shadows, self-avoiding curves, and colored chord diagrams obtained for
n = 2, 3, 4.
{{1, 3}, {2, 5}, {4, 7}, {6, 8}}, {{1, 3}, {2, 6}, {4, 8}, {5, 7}}
{{1, 4}, {2, 7}, {3, 6}, {5, 8}}
For n = 5, the seven chord diagrams are given in the following table:
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
Fig. 3.47 KL shadows, self-avoiding curves, and colored chord diagrams obtained for
n = 5.
Fig. 3.48 (a) Uncolored chord diagram and its bicoloring; (b-e) reconstruction of its
corresponding self-avoiding curve; (f) the corresponding KL shadow.
{{1, 3}, {2, 5}, {4, 7}, {6, 9}, {8, 10}} {{1, 3}, {2, 5}, {4, 8}, {6, 10}, {7, 9}}
{{1, 3}, {2, 5}, {4, 9}, {6, 8}, {7, 10}} {{1, 3}, {2, 6}, {4, 9}, {5, 7}, {8, 10}}
{{1, 3}, {2, 6}, {4, 9}, {5, 8}, {7, 10}} {{1, 3}, {2, 7}, {4, 10}, {5, 9}, {6, 8}}
{{1, 4}, {2, 8}, {3, 7}, {5, 10}, {6, 9}}
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
424 LinKnot
Fig. 3.49 Chord diagrams and self-avoiding curves corresponding to prime KLs for
n = 6.
Fig. 3.50 Three pairs of equal self-avoiding curves (a) shown as shapes (b) and their
chord diagrams (c).
426 LinKnot
Fig. 3.52 Via tori that can be obtained by identifying opposite sides of the rectangle.
Among the chemical elements, carbon C is the basis of all life. A whole
branch of chemistry, organic chemistry, is devoted to the study of C-C
bonds and different molecules originating from them. Carbon is the only
known 4-valent element able to produce long homoatomic stable chains or
different 4-valent nets. Another candidate is silicon, whose homoatomic
chemistry is rapidly developing.
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
Theorem 3.3. Every fullerene has exactly 12 pentagonal faces. For every
even n ≥ 24 there exists at least one fullerene Cn (Grünbaum and Motzkin,
1963; Voytekhovsky and Stepenshchikov, 2005).
428 LinKnot
Fig. 3.53 (a) The four possible vertex configurations; (b) the addition of carbon atom(s)
between two others connected by a double bond; (c) the introduction of bigons in 1111
vertices.
(1) edge-coloring of a 3-regular graph, with one colored edge in each vertex;
(2) introducing bigons in every vertex of a 4-regular graph.
This provides a double check of the obtained results. The duality of these
methods is illustrated in the example of two C20 chemical isomers, both
derived from the same geometrical dodecahedral form with G = [3, 5] =
Ih = S5 of order 120. However, the first has G′ = D5d = [2+, 10] =
D5 × C2 of order 20, and the other G′ = [2, 2]+ = D2 of order 4 (Fig.
3.54a,b). In this case, the symmetry of chemical isomers derived by the
vertex bifurcation is preserved from their generating basic polyhedra (Fig.
3.54b).
For the enumeration of general fullerenes (i.e., source links derived from
basic polyhedra in the Section 2.5) we used the Polya Enumeration Theorem
(PET), applied to basic polyhedra, knowing their automorphism groups (see
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
430 LinKnot
Fig. 3.54 (a) Two non-isomorphic edge colorings of the regular dodecahedron; (b) 5/6
fullerene derived from 10∗ ; (c) 5/6 fullerene derived from 10∗∗ .
Section 2.5), but its application to 5/6 fullerenes is not possible. The same
restriction holds for the other derivation method, because of the condition
that in every vertex exactly one edge of a 3-regular graph must be colored.
The 3-valent graphs with n < 13 vertices and their edge-colorings producing
4-valent graphs are considered by A.Yu. Vesnin (1991).
Similarly, we can prove that 5/6 fullerenes with 22 atoms can not exist,
and there are seven 5/6 fullerenes C24 with the same geometrical form
and G = D6d = [2+, 12] = D12 (Fig. 3.55). Often, chemical symmetry
group G′ is not sufficient for distinguishing chemical isomers. They can be
distinguished using polynomial invariants of KL projections (see Section
2.8).
G = {{1, 2}, {2, 3}, {3, 4}, {4, 5}, {1, 5}, {6, 7}, {7, 8}, {7, 8}, {8, 9}, {9, 10},
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
Fig. 3.55 Seven 5/6 fullerenes C24 with the same geometrical form.
{10, 11}, {10, 11}, {11, 12}, {12, 13}, {13, 14}, {14, 15}, {15, 16}, {16, 17},
{16, 17}, {19, 20}, {17, 18}, {18, 19}, {19, 20}, {6, 20}, {20, 21}, {21, 22},
{22, 23}, {23, 24}, {24, 25}, {25, 26}, {26, 27}, {27, 28}, {28, 29}, {29, 30},
{30, 31}, {31, 32}, {32, 33}, {33, 34}, {34, 35}, {35, 36}, {36, 37}, {37, 38},
{38, 39}, {39, 40}, {21, 40}, {7, 22}, {8, 25}, {10, 26}, {11, 29}, {13, 30},
{14, 33}, {16, 34}, {17, 37}, {19, 38}, {40, 41}, {41, 42}, {42, 43}, {43, 44},
{44, 45}, {45, 46}, {46, 47}, {47, 48}, {48, 49}, {49, 50}, {50, 51}, {51, 52},
{52, 53}, {53, 54}, {54, 55}, {41, 55}, {24, 44}, {27, 45}, {28, 47}, {31, 48},
{32, 50}, {35, 51}, {36, 53}, {39, 54}, {55, 56}, {56, 57}, {57, 58}, {58, 59},
{59, 60}, {56, 60}, {13, 14}, {1, 9}, {1, 9}, {2, 12}, {2, 12}, {3, 15}, {3, 15},
{4, 18}, {4, 18}, {5, 6}, {5, 6}, {21, 40}, {22, 23}, {24, 25}, {26, 27}, {28, 29},
{30, 31}, {32, 33}, {34, 35}, {36, 37}, {38, 39}, {41, 42}, {44, 45}, {47, 48},
{50, 51}, {53, 54}, {55, 56}, {43, 57}, {43, 57}, {46, 58}, {46, 58}, {49, 59},
{49, 59}, {52, 60}, {52, 60}, {23, 42}}
of the fullerene C60 , we obtain the Dowker code of the corresponding link
in the Knotscape format
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
432 LinKnot
dD ′ (t) = t20 − 10t18 + 45t16 − 120t14 + 200t12 − 197t10 + 105t8 − 40t6 + 25t4 − 10t2 ,
dD ′′ (t) = t20 − 10t18 + 45t16 − 120t14 + 208t12 − 250t10 + 217t8 − 130t6 + 49t4 − 10t2 ,
proving their difference. Using the same multivariable invariant for link
projections, we can distinguish seven non-isomorphic diagrams obtained
from the fullerene C24 (Fig. 3.55). The same results can be obtained using
the Liang polynomial.
All 4-valent (chemical) Schlegel diagrams of fullerenes can be converted
into alternating KL diagrams. For example, two chemical isomers of C20
will give knots, and from 7 isomers of C24 we obtain four knots, one 3-
component, one 4-component and one 5-component link. Among the links
obtained, two of them (3-component and 5-component link) contain a min-
imal possible component: hexagonal carbon ring. Notice that C60 consists
only of regularly arranged hexagonal carbon rings, which is maybe the ad-
ditional reason for its stability (Fig. 3.56). Therefore, it will be interesting
to consider the infinite class of 5/6 fullerenes with this property, called per-
fect fullerenes. Some perfect fullerenes were modelled with hexastrips6 by
P. Gerdes (1998). Similar structures, buckling patterns of shells and spher-
ical honeycomb structures have been considered by different authors (e.g.,
T. Tarnai (1989)).
5 Schlegeldiagrams with bigons denoting double bonds.
6 “Framed tangle” from Gauss’ notebook represents a weaving with hexastrips (Przy-
tycki, 2004, Fig. 3.5).
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
Fig. 3.56 Derivation of C60 from C20 and carbon rings in C60 .
Let us describe how to obtain perfect fullerenes from any 5/6 fullerene.
Given 5/6 fullerene in geometrical form (i.e., by a 3-valent graph), apply
mid-edge-truncation and vertex bifurcation in all vertices of the obtained
triangular faces, transforming them into hexagons with alternating bigonal
edges. For example, from C20 , connecting the midpoints of all adjacent
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
434 LinKnot
edges we obtain the 3/5 fullerene covered by connected triangular net and
pentagonal faces preserved from C20 . Then we place bigons in all vertices
of the truncated polyhedron, to turn all triangles into hexagonal faces. In
this way, we derive C60 (in its chemical form) from C20 (Fig. 3.56).
Mid-edge-truncation can be applied to any 5/6 (geometrical) fullerene,
giving a new perfect (chemical) fullerene formed by carbon rings. Simi-
larly, from a 5/6 fullerene with v vertices we can always derive new perfect
5/6 fullerenes with 3v vertices (Fig. 3.58). Moreover, symmetry of the
generating fullerene is preserved. According to the theorem by Grünbaum
and Motzkin (1963), for every non-negative n6 6= 1, there exists a 3-valent
convex 5/6 polyhedron having n5 = 12 pentagonal and n6 hexagonal faces.
Hence, from the infinite class of 3-valent 5/6 polyhedra with v = 20 + n6
vertices, we obtain the infinite class of perfect fullerenes with v = 60 + 3n6
vertices. Perfect fullerenes satisfy two important stability conditions:
The IPR rule means that there are no adjacent pentagons, and HPR
means that all pentagons are “holes”, i.e., every pentagon has only external
double bonds. The first 5/6 fullerene satisfying IPR is C60 , and it also
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
satisfies HPR. The IPR is well known as the stability criterion: all fullerenes
of lower order (less than 60) are unstable, because they do not satisfy IPR.
On the other hand, C70 satisfies IPR, but not HPR (Fig. 3.58).
The same holds for C80 (Fig. 3.58), which has the same icosahedral geo-
metrical symmetry as C60 , but since HPR can not be satisfied, its symmetry
will be reduced due to edge-coloring. Therefore we conclude that only per-
fect fullerenes with G = G′ = [3, 5] = Ih = S5 , satisfying both IPR and
HPR, are C60 , C180 , C240 , etc. We need also to notice that for n6 = 0, 2, 3
there are always exactly one 3-valent 5/6 polyhedron (i.e., the geometrical
form of C20 , C24 , C26 ), but for some larger values (e.g. n6 = 4, 5, 7, 9)
there are several geometrical isomers of generating fullerenes, and conse-
quently, the same number of derived perfect fullerenes (Fig. 3.59). Hence,
considering fullerene isomers, we can distinguish geometrical isomers, that
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
436 LinKnot
438 LinKnot
different C70 isomers with the same geometrical structure (Fig. 3.61) and
the same G and G′ , shows that symmetry is not sufficient for distinguishing
fullerene isomers, so we need additional tools (see the Subsection 3.2.3). In
the same way, from 4-valent graphs with two hexagonal bases, 12 triangular
and 6(2k + 1) quadrilateral faces (k = 0, 1, 2, . . .) we obtain the infinite class
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
of fullerenes C36 , C60 , C84 ,. . . with the symmetry group G = G′ = D6h (Fig.
3.63).
The next series of symmetry groups [2+, 2q] (Dqd ) with q = 5, 6 can
be obtained in the same way, from 4-valent graphs with q-gonal bases, 2q
triangular and 2kq quadrilateral faces (k = 1, 2, . . . for q = 5; k = 0, 1, 2, . . .
for q = 6) (Fig. 3.64). As the limiting case, for q = 5 and k = 0, we obtain
C20 with the icosahedral symmetry group G and G′ = D5d . C20 can be
used as a building block of the whole class of nanotubes C40 , C60 , C80 ,. . .
with G = D5d , (Fig. 3.65). Nanotubes C48 , C72 , C96 ,. . . can be obtained
in the same way, by “gluing” the pentagonal bases, from the fullerene C24
(q = 6, k = 0) (Fig. 3.65).
The geometrical structure of the nanotube class with G = Dqd (q = 5, 6)
permits the edge coloring that preserves symmetry, so there always exist
isomers with G = G′ .
If 3-rotation axis contains the opposite vertices of a fullerene, we have
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
440 LinKnot
442 LinKnot
Fig. 3.71 The infinite perfect 6/7 fullerene in H 2 with heptagonal holes.
444 LinKnot
Definition 3.7. Let us denote the empty set by a line segment − . A finite
expression E in line segments is well-formed if:
Two finite ordered multi-sets are equal iff they have the same members
in the same order, i.e., iff their 0-1 encodings are identical. An isomorphic
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
−
−
−
− − − −
A= B= A= A= -...
The simplest recursive form F = F , beginning from F = − , results in
the sequence of natural numbers − , − , − , − , . . . Recursive form F = F F
can be called Fibonacci form because the number of line segments at depth
n is nth Fibonacci number.
If F (n) denotes the number of line segments (or number of nodes in the
corresponding rooted tree) at depth n of the form F , for any two forms F1
and F2 holds:
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
446 LinKnot
Fig. 3.74 (a) Line segment interpretation of Fibonacci form; (b) rooted tree as its dual.
Instead of using classical Boolean algebra with only two discrete values
0 and 1, we can consider more general algebra whose values belong to
the continuous set [0, 1], and construct a new model of polyvalent Boolean
algebra called the square-free polynomial model. We define p ∧ q and ¬p in
the following way: p ∧ q = pq, ¬p = 1 − p.
Definition 3.10. A square-free polynomial is any polynomial in which a
degree of each variable is 1.
Introducing the idempotent law p2 = p for all variables, we model a
polyvalent Boolean algebra on the continuous interval [0, 1], with square-
free polynomials, standard polynomial multiplication, and idempotency.
This model is very similar to a Boolean ring, but ¬p is defined as ¬p = 1−p,
and not as ¬p = 1 + p. In this way the structure of standard truth tables
is preserved and extended to the set [0, 1].
A concept of square-free polynomials is implicitly given in the original
works of G. Boole (2003).
De Morgan laws imply p ∨ q = p + q − pq, and we can check if our
model is consistent with regard to the whole set of axioms. The following
equalities are obtained by straightforward computations:
a ∧ a = a2 = a, a ∨ a = a + a − a2 = a + a − a = a (idempotency),
a ∧ b = ab = ba = b ∧ a, a ∨ b = a + b − ab = b + a − ba = b ∨ a
(commutativity),
(a ∧ b) ∧ c = (ab)c = a(bc) = a ∧ (b ∧ c), (a ∨ b) ∨ c = (a + b − ab) + c −
(a + b − ab)c = a + b − ab + c − ac − bc + abc = a + b + c − bc − ab − ac + abc =
a + (b + c − bc) − a(b + c − bc) = a ∨ (b ∨ c) (associativity),
a ∨ (a ∧ b) = a + ab − a2b = a + ab − ab = a, a ∧ (a ∨ b) = a(a + b − ab) =
a + ab − a2 b = a + ab − ab = a (absorption),
2
448 LinKnot
− −
− − − − − − −
e= −− (1) = − (2) = − − (2) = − (2) = − (2) =
Denoting ¬p by p, and p ∧ q by pq, using the line segment notation
and square-free polynomials, and interpreting unmarked state as 0, various
tautologies are almost obvious. For example,
p p = (1 − p)p = p − p2 = p − p = 0,
pr qr = 1 − (1 − pr)(1 − qr) = pr + qr − pqr2 = pr + qr − pqr =
(p + q − pq)r = (1 − (1 − p)(1 − q))r = p qr.
The proposed method is very powerful, used even in proofs of more
complex tautologies, e.g., a crosstransposition
q r p r x r y r = r pq rxy .
The left side gives:
qrprxryr =
1 − (1 − (1 − q)(1 − r))(1 − (1 − p)(1 − r))(1 − (1 − x)r)(1 − (1 − y)r) =
1 − pq − pr − qr + 4pqr − r2 + 3pr2 + 3qr2 − 6pqr2 + 2r3 − 3pr3 − 3qr3 +
4pqr3 − r4 + pr4 + qr4 − pqr4 − pqrx − pr2 x − qr2 x + 3pqr2 x − r3 x + 2pr3 x +
2qr3 x − 3pqr3 x + r4 x − pr4 x − qr4 x + pqr4 x − pqry − pr2 y − qr2 y + 3pqr2 y −
r3 y + 2pr3 y + 2qr3 y − 3pqr3 y + r4 y − pr4 y − qr4 y + pqr4 y − pqr2 xy − pr3 xy −
qr3 xy + 2pqr3 xy − r3 xy + pr4 xy + qr4 xy − pqr4 xy =
1 − pq + pqr − rxy.
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
3.5 Waveforms
450 LinKnot
x ∈ [0, 1]. This operator produces a stable state for every y = 1 − x, and a
periodic sequence with the period 2 otherwise.
L.H. Kauffman and F.J. Varela (1980) described an algorithm for pro-
ducing recursive operators which generate periodic sequences of a desired
period. Let a periodic sequence of period p with n variables be given, where
2n−2 < p ≤ 2n . The condition 2n−2 < p implies no variable is constant.
Usually, we are trying to express all periodic sequences with a minimal
number of variables n. A period of such a sequence can be represented in
the form of an n × p array. For example, the sequence of period p = 5 with
n = 3 variables considered before, can be represented as
xyz
010
110
101
011
111
For every 0 in h-column we are associate its preceding row to h, where
h is a variable denoting the head of a column (h = x, y, z). Notice that
the last column is the preceding for the first, since we are working with
cyclic order. In the x-column, 0 appears in the first and fourth row, so the
fifth and third row will be associated to x, i.e., x = (1 1 1)(1 0 1). In the
y-column, 0 appears in the third row, so we associate the second row to y
and obtain y = (1 1 0). In the z-column 0 appears in the first and second
row, so we associate the last and first row to z and obtain z = (1 1 1)(0 1 0).
Replacing every 1 by h, and every 0 by h, and overlining each n-tuple,
we obtain x = xyz xyz, y = xyz, z = xyz xyz, which gives the recursive
operator T (x, y, z) = (xyz xyz, xyz, xyz xyz). Using the square-free poly-
nomial method, we reduce obtained terms, and look for fixed points. Hence,
T (x, y, z) = (xyz xyz, xyz, xyz xyz) = T (1√− xz, √ 1 − xy + xyz, 1 − y + xy +
√
yz − 2xyz). Its fixed point (x, y, z) = ( 3+7 2 , 3−2 2 , 2 − 2) which belongs
to [0, 1]3 we obtain in the same way as before, by solving the system of
equations x = 1 − xz, y = 1 − xy + xyz, z = 1 − y + xy + yz − 2xyz.
Now that we have an algorithm for deriving all sequences with a
given period p, we need tools for distinguishing different sequences: non-
isomorphic, not mutually reverse, and not mutually dual sequences. It
is obvious that two sequences given by a cyclic permutation of rows in
the array will be equal. Also, sequences that are the same up to a per-
mutation of variables can be identified. Two dual sequences, where one
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
452 LinKnot
can be obtained from the other by replacing 0 by 1 and vice versa are
equivalent as well. Cyclically equivalent sequences will have the same
fixed points; sequences with permuted variables will have fixed points per-
muted in the same way; dual sequences will have dual fixed points from
the set [0, 1]n , where n is the number of variables. Representatives of
equivalence classes of periodic sequences are called basic sequences. For
p = 2 and n = 1 we have one basic sequence 0,1,0,1. . . defined by the
operator T (x) = x, with no fixed points on a discrete set {0, 1} and
one fixed point x = 12 on the interval [0, 1]. This sequence embodies
the Liar paradox. For p = 2 and n = 2 we obtain two basic sequences:
(0,0),(1,1). . ., defined by the operator T (x, y) = (xy, xy) = (1 − xy, 1 − xy),
with the fixed point (x, y) = (ϕ, ϕ) on [0, 1]2 , and (0,1),(1,0). . ., defined
by T (x, y) = (xy, xy) = (1 − x + xy, 1 − y + xy), with the fixed point
(x, y) = (1, 1).
For p = 3, the minimal number of variables n is n = 2. From
the set (0,0), (0,1), (1,0), (1,1) we can produce 24 different sequences
of period 3. Each of them satisfies the necessary condition that no
variable is constant. Taking one representative from each equivalence
class, we obtain three basic sequences of period 3: (0,0),(0,1),(1,0). . .,
(0,0),(0,1),(1,1). . ., and (0,0),(1,1),(0,1). . . The first is defined by the op-
erator T (x, y) = (x y xy, xy xy) = (y, xy xy) = T (y, 1 − x − y + 2xy),
the second by T (x, y) = (xy x y, xy) = T (x + y − 2xy, 1 − xy), and
the third by T (x, y) = xy xy = y xy. The fixed points of the first are
(x, y) = (1, 1) and (x, y) = ( 12 , 12 ), of the second (x, y) = ( 12 , 23 ), and of the
third (x, y) = (1 − ϕ, ϕ).
For p = 4, the minimal number of variables n is n = 2, and we
obtain three basic sequences: (0,0),(0,1),(1,0),(1,1). . . with T (x, y) =
(xy xy, xy xy) = (xy xy, y) = (x + y − 2xy, 1 − y); (0,0),(0,1),(1,1),(1,0). . .
with T (x, y) = (x y xy, xy xy) = (y, x) = (y, 1 − x); (0,0),(1,1),(0,1),(1,0). . .
with T (x, y) = (xy xy, xy xy) = (x, xy xy) = (1 − x, 1 − x − y + 2xy). All
of them can not be stabilized for any pair of values from the discrete set
{0, 1}2, and in [0, 1]2 they have the same fixed point ( 12 , 12 ). The same line
of reasoning can be applied to basic periodic sequences with higher periods
and compute their number. For example, for n = 3 there are 45 sequences
of the period p = 4, 160 for p = 5, 382 for p = 6, 840 for p = 7, 840 for
p = 8, etc.
T 2 (x) = xm n m n = xm n = T (x) can be obtained by iterating from
an infinite self-referential form T (x) = xm n with two fixed parameters m
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
454 LinKnot
Fig. 3.77 Reidemeister moves translated to the language of knot automata (Kauffman,
1994).
crete set {0, 1}. In general, if we take values from continuous interval [0, 1],
all the values (x, y) = (x, 1 − x) (x ∈ [0, 1]) give balanced states of this
circuit. A transition consists of reassigning the value of z for the outgoing
edges z of one inverter that is unbalanced. Transition may or may not
result in a balanced state (Kauffman, 1994).
The equation x = x describes the circuit that embodies the Liar para-
dox. In bivalent Boolean logic, it has no solutions for a balanced state, so
it demands a polyvalent logic in order to achieve a stable state. Such a
solution is x = 12 . In a similar way, the automaton defined by equations
x = y, y = x y (Fig. 3.79) has no stable states for the discrete values from
the set {0, 1}, yet stabilizes for (x, y) = (1 − ϕ, ϕ) from the continuous set
[0, 1]2 .
456 LinKnot
Every circuit that has more then one stable state in the discrete set of
values {0, 1}, has a continuum of stable states (i.e., fixed points) in [0, 1],
defined by resulting parametric equations. On the other hand, a circuit
with no stable states in the set {0, 1}, always has a stable state in the
extended set [0, 1], i.e., in polyvalent logic.
The principle of duality holds for automata with NAND and NOR gates:
a system remains unchanged with regard to stable states if all NAND gates
are replaced by NOR gates or vice versa. This holds also for automata that
contain NAND and AND, or NOR and OR gates.
The LinKnot function fAutoSigInp calculates stable states of an au-
tomaton given by a list of outgoing edges, signs of vertices, and inputs in
vertices. For vertices with the sign 1 the operation NAND is used, while for
vertices with sign −1 the operation AND. The result is a list of edge color-
ings corresponding to stable states and a list of stable states according to
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
gates (see Kauffman, 1994), with values taken from the discrete set {0, 1}.
If the list of signs is empty, it is treated as (1, ..., 1), and computations are
made with NAND gates in all vertices of a graph. Inputs in vertices can
be included by giving the corresponding list of vertices. Because only an
input 0 produces a change in NAND or AND gates, all inputs are treated
as 0.
The function fAutoKL calculates stable states of an automaton ob-
tained from a KL given in Conway notation, followed by a list of signs of
vertices, and inputs in vertices.
The logical operation NAND is used in vertices with the sign 1, and
AND in vertices with the sign −1. The result is the oriented graph cor-
responding to a given KL, the list of edge colorings corresponding to its
stable states, and the list of stable states ordered according to gates (see
Kauffman, 1994). If the list of signs of a given KL is empty, the original list
of the signs of a given KL is used for the computation. Inputs in crossings
are the same as for the function fAutoSigInp.
The function fAutoKL can be used for analyzing the behavior of KLs
with regard to stable states, this means, their edge colorings with two col-
ors 0 and 1, compatible with the requirements of NAND and AND logical
gates. For example, an automaton obtained from the figure-eight knot 2 2
(or 41 ) with the signs 1 in all crossings (i.e., with NAND gates in all cross-
ings), defined by equations x = yz, y = xu, z = yu, u = xz, has no stable
states in the discrete set {0, 1}. The corresponding operator T (x, y, z, u) =
(yz, xu, yu, xz) generates the periodic sequence (0,0,0,0),(1,1,1,1). . . for
(x, y, z, u) ∈ {(1, 1, 1, 1), (1, 1, 0, 0), (0, 0, 1, 1)} and for every other initial
August 29, 2007 16:40 World Scientific Book - 9in x 6in ws-book9x6
458 LinKnot
state (x, y, z, u) with more than two zeros. In all other cases, it gener-
ates the sequence of the period eight (0,1,1,0),(1,1,1,1),(0,1,0,1),(1,1,0,1),
(1,0,0,1),(1,0,1,1),(1,0,1,0),(1,1,1,0). . . Using the original signs of the knot
2 2 (this means, NAND gates in the crossings with the sign 1, and AND
gates in the crossings with the sign −1), we obtain the stable state
(x, y, z, u) = (1, 1, 0, 0) (Fig. 3.81a). The mirror image −2 − 2 of the same
knot gives the same result, showing that the figure-eight knot is achiral.
For a figure-eight knot with all signs equal to 1, we have the stable state
(x, y, z, u) = (ϕ, ϕ, ϕ, ϕ) on the set [0, 1]4 (Fig. 3.81b), where ϕ denotes
golden ratio. A figure-eight knot with the signs −1 in all vertices (this
means, with AND gates in all vertices) has two stable states, where all
edges are labelled by 0, or all edges are labelled by 1. On the other hand,
a trefoil 3 has no stable states with NAND gates in all crossings, but its
mirror image −3 with AND gates in all crossings has two stable states from
the discrete set {0, 1}, where all edges are colored by 0, or by 1. This shows
that the trefoil knot is chiral.
July 29, 2007 19:40 World Scientific Book - 9in x 6in bibliographyfinal
Bibliography
459
July 29, 2007 19:40 World Scientific Book - 9in x 6in bibliographyfinal
460 LinKnot
Atiyah, M.F. (1990) The Geometry and Physics of Knots, Cambridge University
Press, Cambridge.
Bain, G. (1973) Celtic Art – the Methods of Construction, Dower, New York.
Bain, I. (1990). Celtic Knotwork, Constable, London.
Balaban, A.T. (1989) Carbon and its nets, in Symmetry: Unifying Human Un-
derstanding 2, (Ed. I. Hargittai), 397–416, Pergamon Press, Oxford, New
York.
Bar Natan, D. (1995) On the Vassiliev Knot Invariants, Topology, 34, 423–472.
Bar Natan, D. (2002) On Khovanov’s categorification of the Jones polynomial,
Algebraic and Geometric Topology, 2, 16, 337–370.
Bar Natan, D. (2006) KnotTheory,
http://katlas.math.toronto.edu/wiki/The Mathematica Package KnotTheory
Barrett, C. (1970) Op-art, Studio Vista, London.
Berge, C. (1985) Graphs, 3rd ed., North-Holland, Amsterdam.
Bernhard, J.A. (1994) Unknotting numbers and their minimal knot diagrams, J.
Knot Theory Ramifications, 3, 1, 1–5.
Bilinski, S. (1985) Die quasiregulären Polyeder vom Geschelcht 2, Österreichische
Akademie der Wissenschaften, Mathematisch-naturwissenschaftliche
Klasse, Band 194, 1–3.
Birkhoff, G.D. (1912) A deretminant formula for the number of ways of coloring
a map, Ann. of Math., 2, 14, 42–46.
Birkhoff, G.D. and Lewis, D.C. (1946) Chromatic polynomials, Trans. Am. Math.
Soc., 60, 355–451.
Birman, J.S. (1976) Braids, Links and Mapping Class Groups, Ann. of Math.
Studies 82, Princeton University Press, Princeton.
Birman, J. (1993) New point of view in knot theory, Bull. Amer. Math. Soc., 28,
253–287.
Bleiler, S.A. (1984) A note on unknotting number, Math. Proc. Camb. Phil. Soc.,
96, 469–471.
Bogle, M.G.V., Hearst, J.E., Jones, V.F.R. and Stoilov, L. (1994) Lissajous Knots,
J. Knot Theory Ramifications, 3, 2, 121–140.
Bohm, J. and Dornberger-Schiff, K. (1966) The nomenclature of crystallographic
symmetry groups, Acta Crystallogr., 21, 1004–1007.
Boo, W.O.J. (1992) An Introduction to Fullerene Structures, J. Chem. Education,
69, 8, 605–609.
Boole, G. (2003) The laws of thought, Prometheus Books, New York.
Boyer, J.M. and Myrvold, W. (2005) On the Cutting Edge: Simplified O(n)
Planarity by Edge Addition, Journal of Graph Algorithms and Applications,
8, 3, 241–273.
Broersma, H.J., Duijvestijn, A.J.W. and Göbel, G. (1993) Generating All 3-
Connected 4-Regular Planar Graphs from the Octahedron Graph, Journal
of Graph Theory, 17, 5, 613–620.
Brunn, H. (1892) Über Verkettung, Sitzungberichte der Bayerischer Akad. Wiss.
Math-Phys., 22, 77–99.
Burde, G. and Zieschang, H. (1985) Knots, Walter de Greyter, Berlin, New York.
July 29, 2007 19:40 World Scientific Book - 9in x 6in bibliographyfinal
Bibliography 461
Calvo, J.A. (1997) Knot Enumeration Through Flypes and Twisted Splices, J.
Knot Theory Ramifications, 6, 6 , 785–798.
Calvo, J.A., Millett, K.C. and Rawdon, E.J. (Eds.) (2002) Physical Knots: Knot-
ting, Linking and Folding Geometric Objects in ℜ3 , AMS Special Session on
Physical Knotting and Unknotting, Las Vegas, Nevada, April 21-22, 2001,
Americal Mathematical Society, Providence, RI.
Caudron, A. (1982) Classification des nœuds et des enlancements, Public. Math.
d’Orsay 82. Univ. Paris Sud, Dept. Math., Orsay.
Cavichiolli, A., Ruini, B. and Spaggiari, F. (2001) On a conjecture of M.J. Dun-
woody, Algebra Colloquium, 8, 2, 169–218.
Cerf, C. (1997) Nullification writhe and chirality of alternating links, J. Knot
Theory Ramifications, 6, 5, 621–632.
Cerf, C. (1998) Atlas of Oriented Knots and Links, Topology Atlas Invited Con-
tributions, 3, 2, 1–32, http://at.yorku.ca/t/a/i/c/31.htm.
Cerf, C. (2001) Topological chirality of knots and links, in Chemical Topology,
Mathematical Chemistry, Vol. 6 (Eds. D. Bonchev and D.H. Roovray), Gor-
don and Breach Publishers, New York, 1–34.
Conant, J. (2006) Chirality and the Conway polynomial,
arXiv:math.GT/0503648v1.
Conway, J. (1970) An enumeration of knots and links and some of their related
properties, in Computational Problems in Abstract Algebra, Proc. Conf.
Oxford 1967 (Ed. J. Leech), 329–358, Pergamon Press, New York.
Conway, J.H. and Gordon, C.McA. (1983) Knots and Links in Spatial Graphs, J.
Graph Theory, 7, 445–453.
Coxeter, H.S.M. (1994) Symmetrical combinations of three and four hollow tri-
angles, Math. Intelligencer, 16, 3, 25–30.
Coxeter, H.S.M. and Moser, W.O.J. (1980) Generators and Relations for Discrete
Groups, Springer-Verlag, New York.
Cromwell, P.R. (1993) Celtic knotwork: mathematical art, The Math. Intelli-
gencer, 15, 1, 36–47.
Cromwell, P. (1995) Borromean triangles in Viking art, Math. Intelligencer, 17,
1, 3–4.
Cromwell, P. (2004) Knots and links, Cambridge University Press, Cambridge.
Cromwell, P., Beltrami, E. and Rampichini, M. (1998) The Borromean rings,
Math. Intelligencer, 20, 53–62.
Crowell, R.H. and Fox R.H. (1965) Knot Theory, Blaisdell Publishing Company,
New York, Toronto, London.
Dabkowski, M.K. and Przytycki, J.H. (2002) Burnside obstructions to the
Montesinos-Nakanishi 3-move conjecture, Geometry and Topology, 6, 335–
360.
Darcy, I.K. and Sumners, D.W. (1998) Applications of toppology to DN A, in
Knot Theory (Eds. V.F.R. Jones et all), Banach Center Publ. 42, Polish
Acad. Sci., Inst. Math., Warszawa.
Darcy, I.K. and Sumners, D.W. (2000) Rational tangle distances of knot and
links, Math. Proc. Cambridge Philos. Soc., 128, 497–510.
July 29, 2007 19:40 World Scientific Book - 9in x 6in bibliographyfinal
462 LinKnot
Bibliography 463
464 LinKnot
Bibliography 465
466 LinKnot
Jones, V.F.R. (1987) Hecke algebra representations of braid groups and link poly-
nomials, Ann. of Math., 126, 335–388.
Jones, V.F.R. (1989) On knot invariants related to some statistical mechanical
models, Pacific J. Math., 137, 311–334.
Jones, V.F.R. (2005) The Jones polynomial,
http://math.berkeley.edu/∼ vfr/jones.pdf
Jordan, M.C. (1887) Cours d’analyse, Tome 3, L’École Polytechnique, Gauthier-
Villars, Paris.
Joyce, D. (1982) A Classifying Invariant of Knots, The Knot Quandle, J. Pure
Appl. Alg. 23, 37–65.
Kanenobu, T. (1986) Infinitely many knots with the same polynomial invariant,
Proc. Amer. Math. Soc., 97, 1, 158–162.
Kanenobu, T. and Murakami, H. (1986) Two-bridge knots with unknotting num-
ber one, Proc. Amer. Math. Soc., 98, 499–502.
Kauffman, L.H. (1983) Combinatorics and knot theory, Contemporary Mathemat-
ics, 20, 181–200.
Kauffman, L.H. (1987a) On Knots, Princeton University Press, Princeton.
Kauffman, L.H. (1987b) State models and the Jones polynomial, Topology, 26,
395–407.
Kauffman, L.H. (1988) New invariants in the theory of knots, Amer. Math. Mon.,
March, 195–242.
Kauffman, L.H. (1990) An invariant of regular isotopy, Trans. Amer. Math. Soc.,
318, 417–471.
Kauffman, L.H. (1990a) Problems in Knot Theory, Open Problems in Topology
(Eds. J. van Mill and G.M. Reed), Elsevier Science Publishers B.V., North
Holland, 487–522.
Kauffman, L.H. (1991) Knots and physics, World Scientific, Singapore.
Kauffman, L.H. (1994) Knot automata, IEEE Computer Society Press Reprint,
328–333.
Kauffman, L.H. (1995) Knot logic, in Knots and Applications (Ed. L. Kauffman),
World Scientific, 1–110.
Kauffman, L.H. (1995a) Hopf algebras and invariants of 3-manifolds, Journal of
Pure and Appl. Algebra, 100, 73–92.
Kauffman, L.H. (1997) Virtual knots, Talks at MSRI Meeting, January 1997 and
AMS Meeting, at University of Maryland, College Park, March 1997.
Kauffman, L.H. (1999) Virtual Knot Theory, European J. Comb., 20, 663–690.
Kauffman, L.H. (2000) A Survey of Virtual Knot Theory, Knots in Hellas ’98,
Proceedings of the International Conference on Knot Theory and its Ram-
ifications (Eds. C.McA. Gordon, V.F.R. Jones, L.H. Kauffman, S. Lam-
bropoulou, J.H. Przytycki), World Scientific, Singapore, New Jersey, Lon-
don, Hong Kong, 143–202.
Kauffman, L.H. (2001) Detecting virtual knots, Atti. Sem. Math. fis., Univ. Mod-
ena, Supplemento al Vol. IL, 241–282.
Kauffman, L.H. (2004) Knot diagrammatics, arXiv:math.GN/0410329v5.
Kauffman, L.H. and Lambropoulou, S. (2003) On the classification of rational
tangles, arXiv:math.GT/0311499v2.
July 29, 2007 19:40 World Scientific Book - 9in x 6in bibliographyfinal
Bibliography 467
468 LinKnot
Bibliography 469
Mackay, A.L. and Terrones, H. (1993) Hypothetical graphite structures with neg-
ative gaussian curvature, Phil. Trans. Roy. Soc. London A, 343, 113–127.
Makarychev, Y. (1997) A Short Proof of Kuratowski’s Graph Planarity Criterion,
The Journal of Graph Theory, 25, 129–131.
Manturov, V.O. (2000a) Bifurcations, atoms and knots, Moscow Univ. Math.
Bul., 1, 3–8.
Manturov, V.O. (2000b) The bracket semigroup of knots, Mathematical Notes,
67, 4, 571–581.
Manturov, V.O. (2002) On invariants of virtual links, Acta Applicandae Mathe-
maticae, 72, 3, 295–309.
Manturov, V.O. (2003) Multivariable polynomial invariants for virtual knots and
links, J. Knot Theory Ramifications, 12, 8, 1131–1144.
Manturov, V.O. (2004) Knot Theory, Chapman and Hall/CRC, Boca Raton,
London, New York, Washington, DC.
Markov, A.A. (1935) Über de freie Aquivalenz geschlossener Zöpfe, Recqueil Math-
ematique Moscou, 1, 73–78.
Markov, A.A. (1945) Foundations of the algebraic theory of braids, Trudy math.
inst. Steklov, 16, 3-54 (in Russian).
Martin, G.E. (1980) Transformation Geometry, Springer-Verlag, New York, Hei-
delberg, Berlin.
Marx, M.L. (1969) The Gauss realizability problem, Proc. Amer. Math. Soc., 22,
610–613.
Matveev, S.V. (1982) Distributivnye gruppoidy v teorii uzlov, Mat. Sbornik, 119,
1, 78–88 (in Russian).
Matveev, S.V. (2003) Algorithmic topology and classification of 3-manifolds,
Springer-Verlag, New York, Berlin, Heidelberg.
Menasco, W.W. and Thistlethwaite, M.B. (1991) The Tait flyping conjecture,
Bull. Amer. Math. Soc., 25, 2, 403–412.
Menasco, W.W. and Thistlethwaite, M.B. (1993) The classification of alternating
links, Annals of Mathematics, 138, 1, 113–171.
Milnor, J.W. (1968) Singular points of complex hypersurfaces, Annals of Mathe-
matics Studies 61, Princeton University Press, Princeton.
Mohar, B. and Thomassen C. (2001) Graphs on Surfaces, Johns Hopkins Univer-
sity Press, Baltimore.
Montesinos, J.M. (1984) Revetements ramifies des nœuds, Espaces fibres de
Seifert et scindaments de Heegaard, Publicaciones del Seminario Mathe-
matico Gercia del Galdeano, Serie II, Seccion 3.
Moriuchi, H. (2004) An enumeration of theta-curves with up to seven crossings,
http://knot.kaist.ac.kr/2004/proceedings/MORIUCHI.pdf
Motegi, K. (1996) A note on unlinking numbers of Montesinos links, Rev. Mat.
Univ. Complut. Madrid, 9, 1, 151–164.
Murakami, H. (1985) Some Metrics on Classical Knots, Math. Ann., 270, 35–45.
Murasugi, K. (1958) On the Alexander polynomial of the alternating knot, Osaka
Math. J., 10, 181–189.
Murasugi, K. (1965) On a certain numerical invariant of link types, Trans. Amer.
Math. Soc., 117, 387–422.
July 29, 2007 19:40 World Scientific Book - 9in x 6in bibliographyfinal
470 LinKnot
Bibliography 471
472 LinKnot
Scharlemann, M. (1985) Unknotting number one knots are prime, Invent. Math.,
82, 37–55.
Schubert, H. (1949), Die eindeutige Zerlegbarkeit eines Knotens in Primknoten,
S.-B. Heidelberger Akad. Wiss. Math.-Nat. Kl., 57–104.
Schubert, H. (1954) Über eine numerische Knoteninvariante, Math. Zeit., 61,
245–288.
Schubert, H. (1956) Knoten mit zwei Brücken, Math. Zeit., 65, 133–170.
Seifert, H. (1934) Über das Geschlecht von Knoten, Math. Ann., 110, 571–592.
Seguin, C.H. (2001) Viae Globi – Pathways on a Sphere, Proc. Mathematics and
Design Conference, Geelong, Australia, July 3-5, 366–374.
Shubnikov, A.V. and Koptsik, V.A. (1974) Symmetry in science and art, Plenum
Press, New York, London.
Shumakovitch, A. (2004) Torsion of the Khovanov homology,
arXiv:math.GT/0405474v1.
Siebenmann, L. (1975) Exercices sur les nœuds rationnels, Orsay (unpublished).
Simony, O. (1882) Über eine Reihe neuer Thatsachen aus dem Gebeite der Topolo-
gie, Math. Annalen, 19, 110–120.
Simony, O. (1884) Über eine Reihe neuer Thatsachen aus dem Gebeite der Topolo-
gie, Math. Annalen, 24, 253–280.
Singmaster, D. (1975) Number of unlabeled Hamiltonian circuits on n-
octahedron, J. Combinatorial Theory, Ser. B, 19, 1, 1–4.
Sossinsky, A. (2002) Knots– Mathematics with Twist, Harvard University Press,
Cambridge (Mass.).
Spencer-Brown, G. (1969) Laws of Form, George Allen and Unwin Ltd., London.
Stanley, R.P. (1986) Enumerative Combinatorics, Wadsworth Inc., Monterey
(California).
Sumners, D.W. (1988) The knot enumeration problem, Studies in Physical and
Theorethical Chemistry, 57, 67–82.
Sumners, D.W. (1990) Untangling DNA, Mathematical Intelligencer, 12, 3, 71–
80.
Sumners, D.W. (Editor) (1993) New Scientific Applications of Geometry and
Topology, Proceedings of Symposia on Applied Mathematics, American
Mathematical Society, Vol. 45.
Stoimenow, A. (2001) Some examples related to 4-genera, unknotting numbers,
and knot polynomials, Jour. London Math. Soc., 63, 2, 487–500.
Stoimenow, A. (2003) On the unknotting number of minimal diagrams, Mathe-
matics of Computation, 72, 244, 2043–2057.
Stoimenow, A. (2004) On unknotting numbers and knot triviadjacency, Mathe-
matica Scandinavica 94, 2, 227–248.
Stoimenow, A. (2005) Square numbers, spanning trees and invariants of achiral
knots, Communications in Analysis and Geometry, 13, 3, 591–631.
Tait, P.G. (1876/77a) On knots, Trans. Roy. Soc. Edin., 9, 97, 306–317.
Tait, P.G. (1876/77b) On links, Trans. Roy. Soc. Edin., 9, 98, 321–332.
Tait, P.G. (1876/77c) On knots I, Trans. Roy. Soc. Edin., 28, 145–190.
Tait, P.G. (1883/84) On knots II, Trans. Roy. Soc. Edin., 32, 327–342.
Tait, P.G. (1884/85) On knots III, Trans. Roy. Soc. Edin., 32, 493–506.
July 29, 2007 19:40 World Scientific Book - 9in x 6in bibliographyfinal
Bibliography 473
474 LinKnot
Index
475
July 29, 2007 19:40 World Scientific Book - 9in x 6in indexfinal
476 LinKnot
Index 477
478 LinKnot
Index 479
480 LinKnot
Index 481
482 LinKnot
Index 483
484 LinKnot
Index 485
walk, 2
closed, 3
open, 3
Whitney degree, 45
wild knot, 11
winding number, 45
Wirtinger presentation, 201
word, 16
world, 50, 59, 195, 304, 381
arborescent, 64, 229