Você está na página 1de 54

Invariants of Knots and Links:

Zeros of the Jones Polynomial


Neil Wright
Supervisor: Prof. E. Corrigan
April 29, 2010
Acknowledgement
I would like to thank my supervisor Prof. E. Corrigan for his advice and
guidance throughout the project. I would also like to thank Liam, Alex and
Neil for their support.
Declaration
This piece of work is a result of my own work except where it forms an
assessment based on group project work. In the case of a group project,
the work has been prepared in collaboration with other members of the
group. Material from the work of others not involved in the project has
been acknowledged and quotations and paraphrases suitably indicated.
Creation of Images
Unless otherwise stated: Images of knots and links, and link diagrams, were
created with [23] (and, where necessary, edited with [24]). Plots were created
with [25]. Graphs and other images were created with [24].
Abstract
Following an introduction to knot theory, we will look at the relationship
between the Jones polynomial and statistical mechanics. This serves as mo-
tivation to study the zeros of the Jones polynomial for families of links. We
will then use graph polynomials to nd general expressions of the Jones
polynomial for some families of links. This will allow us to nd the accumu-
lation sets of the Jones polynomial zeros as the number of crossings in the
links tends to innity.
Contents
1 Introduction to Knots and Links 2
2 The Jones Polynomial 9
3 The Tutte and Chain Polynomials of Graphs 16
4 Torus Links 22
5 Pretzel Links 24
6 Montesinos Links 28
7 Zeros of the Jones Polynomial 34
8 Conclusion 41
Bibliography 42
A A Knot Table 44
B Jones Polynomial Results 45
C Chain Polynomial Results 48
1
Chapter 1
Introduction to Knots and
Links
Figure 1.1: A knot Figure 1.2: A link
Knot theory is the study of the ways in which we can tie a piece string, or
how a 1 dimensional string can lie in ordinary three-dimensional space
[2]. In the late 19th century, a theory of Lord Kelvin was that atoms were
knotted, and so the properties of elements were related to the way in which
the atoms were knotted [1]. This led in 1877 to P.G. Tait beginning the
enumeration of knots: obtaining a table of knots such that knots which
were equivalent (i.e. could be deformed into each other) only appear in the
table once [1]. See Appendix A for an example of a knot table.
The theory of invariants of knots and links is to nd ways to establish
whether two given knots are equivalent, or are in fact distinct knots. The
primary aim of research is to nd better invariants that are able to distin-
guish more knots, but there are other interesting aspects to knot invariants
such as their relations to mathematics in other areas. In this report, we will
later concentrate on the relation between one particular invariant, the Jones
polynomial, and the Potts model used in statistical mechnics.
2
1.1 Denition
In the denition of knots (and links) we want to capture the idea of a
knotted loop (or in the case of links, loops) of rope. Essentially, a knot is an
embedding of a circle, S
2
, into R
3
, but we wish to avoid some complications.
Firstly, we want to avoid self-intersections in knots and links. Secondly we
want to avoid wild knots and links, where a knot or link contains an innite
number of kinks getting progressively smaller, as shown in Figure 1.3.
Figure 1.3: A wild knot
By dening knots and links via piecewise linear closed curves we can avoid
these complications:
Denition 1.1 [2]. A link of m components is a subset of S
3
, or of R
3
, that
consists of m disjoint, piecewise linear, simple closed curves. A link of one
component is a knot.
Although piecewise linear implies that the curves are made up of straight
segments, we generally assume there are so many of these segments that
images and diagrams of knots and links are drawn with smoothly curved
components [2]. Since the denition of links includes knots as a subset, we
will generally use the word link when referring to both (and only use knot
when we want to refer specically to one component links).
The unknot is the simplest knot, and is just the unkotted circle [3] (see
Figure 1.4).
Figure 1.4: The unknot
We can give links the addition structure of orientation. An oriented link is
a link where each component is given a direction (i.e. each component is
the map of an oriented circle [4]).
3
1.2 Equivalence
Two links are considered intuitively equivalent if we can deform one into the
other without passing through itself and without making parts dissappear
by pulling them tighter and tighter. We dene equivalence via homeomor-
phisms on the whole of S
3
to capture this idea and avoid any complications:
Denition 1.2 [2]. Links L
1
and L
2
in S
3
are equivalent if there is an
orientation preserving piecewise linear homeomorphism h : S
3
S
3
such
that h(L
1
) = (L
2
).
Links that are equivalent can be regarded as being simply the same link.
1.3 Diagrams
To show and study links we use link diagrams. A link diagram is a projection
of a link in R
3
onto R
2
. Where the knot passes over/under itself, we get a
crossing in the diagram like the one shown in Figure 1.5. The break is one
segment (the underpass) shows that it passes behind the other (the overpass)
as we view it. In link diagrams we want to avoid any ambiguity, so we would
not allow crossings like the one shown in Figure 1.6. If necessary, we can
show orientation of link components in a link diagram by adding arrowheads
to path components. An example in Figure 1.7 shows an unoriented knot
called the trefoil knot and a possible diagram of it.
overpass underpass
Figure 1.5: A crossing, used in a
link diagram
Figure 1.6: An ambiguous cross-
ing, which we would not use in a
link diagram
Figure 1.7: The trefoil knot and a diagram of the trefoil knot
4
One property of a link diagram that will be used in later denitions is that
of writhe:
Denition 1.3 [2]. The writhe w(D) of a diagram D of an oriented link is
the sum of the signs of the crossings of D, where each crossing has sign +1
or 1 as dened (by convention) in Figure 1.8.
+1 -1
Figure 1.8: The sign of a crossing in a link diagram
1.3.1 Reidemester moves
If we have two dierent diagrams of the same link (i.e. equivalent links),
then the diagrams are related by the three Reidemeister moves and an
orientation-preserving homeomorphism of the plane [2]. The orientation-
preserving homeomorphism of the plane amounts to warping or changing
the link diagram (or parts of the link diagram) without altering the cross-
ings. The three types of Reidemeister move are show in Figures 1.9, 1.10
and 1.11.

Figure 1.9: Type I Reidemeister


move

Figure 1.10: Type II Reidemeister


move

Figure 1.11: Type III Reidemeister move


5
1.4 Prime knots
We can dene the composition of knots as follows: Remove a small arc from
each diagram of two knots and then join the loose ends [3] (see Figure
1.12). The composition of two knots K and J may be denoted by K#J (as
in [3]) or by K+J (as in [2]). The composition of any knot with the unknot
will just give that knot [3].
K
J K # J
Figure 1.12: Composition of knots
A composite knot is a knot that can be formed by the composition of two
other knots (which are not the unknot). A knot that cannot be made in this
way is called a prime knot [3]. Appendix A shows the prime knots with up
to 7 crossings.
If we have two oriented knots K and J then there are two ways to compose
them: with their orientations corresponding (see Figure 1.13a); or with their
orientations not corresponding (see Figure 1.13b). In the rst case, the knot
K#J will be the same wherever we compose the knots. This is also true in
the seond case, but the knots formed in each case may not be the same [3].
K J K # J
(a) Corresponding orientation
K J K # J
(b) Opposite orientation
Figure 1.13: Composition of oriented knots
1.5 Invariants
The fundamental problem in knot theory is to distinguish links that are not
equivalent, and to nd when two (perhaps very dierent) diagrams actually
represent the same link. A link invariant assigns to a link a mathmetical
object (e.g. a number, a polynomial, a group) and gives the same result for
equivalent links [2]. Equivalently, it will not change under the actions of the
three Reidemeister moves [1].
6
1.5.1 Examples
The are many dierent link invariants, here we present some examples of
the most commonly seen invariants.
Crossing number
The crossing number (or crossing index [1]) of a link is the minimum number
of crossings used in any diagram of the link [2]. The unkot has crossing
number zero, while there is no knot with crossing number 1 (because any
way we join the four ends of a single crossing we obtain a diagram of a knot
equivalent to the unkot) [3].
Unknotting number
The unknotting number of a link is the least number of crossings that need
to be changed (i.e. the overpass and underpass swapped) in any diagram of
the link to obtain a diagram of the unknot [1].
Colourability
A knot is colourable (or tricolourable) if we can assign one of three colours
to each arc in the knot diagram, such that [1]:
1. At least 2 colours are used.
2. At any crossing where 2 colours appear, all three colours appear.
(It does not matter which diagram of a knot we use, since if a knot is
colourable all diagrams of the knot will be colourable [1].) The unkot
(see Figure 1.4) is not colourable, but the trefoil knot (see Figure 1.7) is
colourable [3].
Mod p Labelling
This is a generalisation of colourability, where the case p = 3 corresponds
to the denition of colourability above [1]. A knot can be mod p labelled if
we can assign to each arc in the knot diagram an integer from 0 to p 1,
such that [1]:
7
1. At least 2 labels are distinct.
2. At every crossing we have
2x y z = 0 (mod p)
where x is the label on the overpass, and y and z are the labels on the
two parts of the underpass.
Genus
We can associate with a knot an orientable surface which has the knot as its
boundary. This surface is called a Seifert surface of the knot (it may not be
unique) [1]. The genus of a knot is the minimum possible genus of a Seifert
surface of the knot.
Group
The group of a link is the fundamental group of the complement of the link
in R
3
[4].
Polynomials
There are a number of link invariants that are polynomials (in one or more
variables). The rst of these was introduced in 1928 by James Alexander,
and is the Alexander polynomial [3]. It is a Laurent polynomial associated
with oriented links [2]. In 1969 John Conway showed that Alexander poly-
nomials could be calculated using only the polynomial of the unknot and
a skein relation (a relation connecting the polynomials of three links whose
diagrams dier at one crossing) [3]. He introduced the Conway polynomial,
which can be obtained from the symmetric Alexander polynomial by a sub-
stitution of variables [1].
Many more polynomials based on such skein relations have been developed,
including the Jones polynomial introduced in 1987 by Vaughan Jones in [5].
The Jones polynomial of an oriented link is a Laurent polynomial in t
1
2
with
integer coecients. In this report we are concerned with the Jones polyno-
mial beacuse of its relationship to the mathematics of statistical mechanics.
The HOMFLY polynomial is a two-variable polynomial generalisation of
the Alexander and Jones polynomials [3].
8
Chapter 2
The Jones Polynomial
V
L
= t +t
3
t
4
V
L
= t
1
2
t
5
2
Figure 2.1: A knot and a link and their Jones polynomials
The Jones polynomial, V (L) or V
L
, of an oriented link L is a Laurent poly-
nomial in t
1
2
with integer coecients. It was introduced by V.F.R Jones in
1987 (see [5]). The Jones polynomial is calculated from the link diagram
and is invariant under the Reidemester moves [2]. Two simple examples of
Jones polynomials are given in Figure 2.1.
Although the Jones polynomial was rst introduced via von Neumann al-
gebras, it can be dened via the bracket polynomial of L.H. Kauman (see
[7]) as follows:
Defnition 2.1 [2]. The Jones polynomial V (L) of an oriented link L is the
Laurent polynomial in t
1
2
, with integer coecients, dened by
V (L) =
_
(A)
3w(D)
D
_
t
1
2 =A
2
Z
_
t

1
2
, t
1
2
_
where D is any oriented diagram for L, w(D) is the writhe of D, and D is
the bracket polynomial of the unoriented diagram.
9
2.1 The Kauman Bracket Polynomial
The Kauman bracket, D, assigns to an unoriented link diagram D a
Laurent polynomial in A with integer coecients [2]. It can be dened by
the three relations
(i) = 1,
(ii) D =
_
A
2
A
2
_
D,
(iii) = A +A
1
,
where the brackets in (iii) contain link diagrams that are identical except in
one place where they dier as shown [2].
Consider how the three Reidemeister moves aect the Kaufmann bracket:
Lemma 2.2 [2]. If a diagram is changed by a Type I Reidemeister move,
its bracket polynomial changes in the following way:
= A
3

= A
3

Lemma 2.3 [2]. If a diagram D is changed by a Type II or Type III Rei-
demeister move, then D does not change.
The writhe of a link diagram will also not be changed by Type II or Type III
Reidemeister moves but a Type I Reidemeister move will change the writhe
by +1 or 1 [2]. In fact (for either possible orientation),
w( ) = w( ) + 1
w( ) = w( ) 1 .
Combining this with Lemmas 2.2 and 2.3 and Denition 2.1, we see that the
Jones polynomial is invariant under all three Reidemeister moves and so is
a link invariant [2].
2.1.1 An Example
An example calculation of a Jones polynomial via the Kauman bracket:
= A +A
1

10
By Lemma 2.2 we have
= (A
3
)(A
3
) = A
6
.
By Lemma 2.2 we also have
=A
3
= A
3
and
=A
3
= A
3
.
So we have
=A +A
1

=A(A
3
) +A
1
(A
3
)
=A
4
A
4
.
So
=A(A
4
A
4
) +A
1
(A
6
)
=A
5
A
3
+A
7
.
From the oriented diagram we have
w
_ _
= 3
So
(A)
3w
=A
9
(A
5
A
3
+A
7
)
=A
4
+A
12
+A
16
V
_ _
=t +t
3
t
4
.
2.2 Properties
The Jones polynomial obeys the following skein relation [2]:
_
t
1
2
t

1
2
_
V
_ _
= t
1
V
_ _
tV
_ _
Where the link diagrams are identical except in one place where they dier
as shown.
If L is an oriented link with Jones polynomial V (L), substituting t
1
for t
gives the Jones polynomial for the mirror image of L [2].
11
If K is an oriented knot formed by the composition of oriented knots K
1
and K
2
, then [2]
V (K) = V (K
1
) V (K
2
) .
2.3 Statistical Mechanics
In introducing the Jones polynomial [5], V.F.R. Jones established a rela-
tionship between knot theory and statistical mechanics that has led to lots
of research [3]. The von Neumann algebra used in [5] was seen to be resem-
ble the algebra of the Temperley-Lieb formulation of the Potts model [16].
The Yang-Baxter equation is another connection between knot theory and
physics, and is now used in knot theory although originating in physics [4].
The parallel between the Star-Triangle relation and Type III Reidemeister
moves is another area of interest in the connection between knot theory and
statistical mechanics [16].
In this report we are interested in the direct relation between the Jones
polynomial and the partition function of the Potts model in statistical me-
chanics (the Jones polynomial is given by a special case of the partition
function [18]). This motivates us to nd general expressions of Jones poly-
nomials for families of links, so that we may nd the distribution of the zeros
of Jones polynomials as the number of crossings tends to inty.
2.3.1 The Potts Model
In statistical mechnics, the Ising model is used to study systems of particles
where only particles close together interact [3] and we wish to study the
overall properties of the system [8]. The Ising model considers particles
situated at vertices on a graph, where each particle is in one of two spins
states and interactions occur along the egdes of the graph (the graphs are
often lattices) [3]. The Potts model generalises the Ising model to allow each
particle be in any one of q states [3].
As in [8], let us consider a simple Potts model where the interaction energy
between particles is constant and only interactions between particles directly
connected by an edge are considered. Given a graph G and a set S of q
elements (the spin states), then assigning an element of S to each vertex
of G we obtain a particular state of G [8]. Two possible Hamiltonians (the
energy of a particular state of G) are [8]
h
1
() = J

i,j
(
i
,
j
) and h
2
() = J

i,j
(1 (
i
,
j
))
12
where is a state of G, i and j are vertices of G in spin states
i
,
j
, is
the Kronecker delta function and J is the interaction energy (if J > 0 the
model is feromagnetic and if J < 0 it is antiferromagnetic [8]). The partition
function of the model (from which we can study the thermodynamics of the
system) is then [8]
Z
i
(G) =

e
(h
i
())
where the sum is taken over all possible . Choosing i = 1 or 2 will give the
respective Hamiltonians. And =
1
kT
, where T is the temperature of the
system and k is the Boltzmann constant [8].
The Jones polynomial can be directly related to the partition function of
the Potts model; an exact relation between the two is given in [14] and
described again here. Firstly, consider the Potts model where we take the
Hamiltonian as h
1
(w) above, and set K = J. Now let K vary so that
each pair of interacting spins, i and j, has an interaction K
i,j
. We get the
following partition function:
Z(G) =

exp
_
_

i,j
K
i,j
(
i
, j)
_
_
=

i,j
e
K
i,j
(
i
,j)
=

i,j
_
1 +
_
e
K
i,j
1
_
(
i
,
j
)
_
where the product is taken over all edges of G. This is exactly the Potts
model partition function given in [14]. The partition function is then a
multivariate polynomial in q and e
K
i,j
.
Now consider an oriented link with Jones Polynomial V (t) and its link di-
agram. Shade alternate regions of the link diagram (there are two possible
shadings, the inverse of each other). Place a particle (i.e. a vertex of the
graph) in each shaded region and place interactions (i.e. edges of the graph)
along crossings. An interaction is K
+
when the crossing is + and K

when
the crossing is , as shown in Figure 1.8. A simple example of this process
is given in Figure 2.3. The Jones polynomial of the link is then related to
the partition function of the Potts model on the graph, Z(q; e
K
+
, e
K

), by
the relation [14]
V (t) = q
(M+1)/2
(t)
N
t
(3n
+
+m
+
)/2
Z
_
q; e
K
+
, e
K

_
where n

is the number of crossings in the shaded link diagram (as shown


in Figure 2.2), m

is the number of K

interactions, N is the total number


of crossings in the link diagram (so N = n
+
+ n

= m
+
+ m

), M is the
13
number of vertices of the graph (i.e. shaded regions of the link diagram),
q = t + 2 +t
1
, e
K
+
= (t)
1
and e
K

= t.
+
-
Figure 2.2: Sign given to crossings of the shaded link diagram
+
+
+
+
+
+
+ +
Figure 2.3: A shaded link diagram and the associated Potts model graph
If we choose the alternate shading on the link diagram, then we will in-
terchange each K
+
and K (so m
+
and m

will also interchange). The


relation will then obtain the same Jones polynomial but with an extra sign
(1)
N
[14].
2.3.2 Zeros
In [12] and [13], Yang and Lee highlighted the study of zeros of partition
functions, in paticular to study phase transitions.
In [12] a monatomic lattice gas in a box of volume V (allowed to exchange
atoms with an external reservior, with a xed temperature), and the grand
partition function are considered. The thermodynamic functions of pressure
and density can be calculated from the partition function, and are given by
the limits of their average values as volume tends to innity [12]. So the
location of the zeros of the grand partition function as V will dictate
where the thermodynamic functions remain analytic:
Their [the zeros of the grand partition function] distribution in
14
the limit V gives the complete analytic behaviour of the
thermodynamic functions... [12]
Phase transitions will only occur where the density function is not analytic,
so only at points where the zeros of the grand partition function converge
as V [12]. (In particular, if we consider the zeros in the complex plane
then the axis corresponding to real, positive values of a physical variable are
of interest. [12])
In [13] it is shown that the study of the Ising model in a magnetic eld is
equivalent to the lattice gas model, so the study of their thermodynamic
properties is equivalent. In particular, the partition function (in the Ising
model) is proportional to the grand partition function (in the lattice gas
model) and the number of spin states corresponds to the volume of gas [13].
Given the importance of zeros of partition functions to thermodynamic prop-
erties and the relation between the Jones polynomial and the Potts model
partition function (given in Section 2.3.1), we are motivated to study the
zeros of Jones polynomials. In particular, we wish to nd accumulation sets
of zeros as the number of crossings in a family of links tends to innity. This
appears to have been intially done by Wu and Wang in [14] and continued
by Chang and Shrock in [15] and then Jin and Zhang in [16], [17] and [18].
15
Chapter 3
The Tutte and Chain
Polynomials of Graphs
To study the zeros of Jones polynomials of links, we rst want to nd general
expressions for the Jones polynomials of families of links. An approach used
in [15] is to use the connection between the Jones polynomial of an alter-
nating link and the Tutte polynomial of a graph. A connection between the
Tutte polynomial and the chain polynomial can be used to nd Tutte poly-
nomials for famlies of related graphs, as used in [16]. This is the approach
we will use, but rst we must establish some properties of the polynomials
and the relations connecting them.
3.1 The Associated Graph of a Link Diagram
The following method of assigning a planar graph to a connected, alternating
link diagram is used in [15] and [16]. (Here we shall continue with the
notation of [16].) A link diagram will divide the plane into a number of
regions. At each crossing of a link diagram, we can assign each of the four
small regions as an A-channel or B-channel as shown in Figure 3.1. We can
A
B
A
A A
B
B
B
Figure 3.1: Assignment of channels to a link diagram
16
then call each region of the link diagram an A-(B-)region if all the channels
it contains are A-(B-)channels. Assign to each A-region a vertex and join
two vertices if they lie either side of a crossing in the link diagram. We shall
call the planar graph that has been created the associated graph of the link
diagram.
In this report we generally use standard diagrams for families of knots, so
we will obtain standard graphs for families of knots. We will denote the
associated graph obtained from the standard diagram of a link L by G
L
.
3.2 The Tutte Polynomial
The Tutte polynomial, T[G](x, y), is a two variable polynomial that we
associate with a graph G. We can dene the Tutte polynomial via the
following rules of deletion and contraction of edges of a graph [8]:
1. If e is an edge of a graph G, and it is neither an isthmus (an edge
which if deleted will increase the number of components of G) or a
loop (an edge with the same vertex at its endpoints), then
T[G] = T[Ge] +T[G/e] .
2. If G consists of i isthmuses and j loops, then
T[G] = x
i
y
j
.
Where Ge is the graph obtained from G by removing the edge e, and G/e
is the graph obtained from G by contracting (i.e. removing and identify the
end vertices of) the edge e. So we can calculate the Tutte polynomial of
a graph by removing and contracting edges until we have a graph of only
isthmuses and loops (the order in which we do these steps does not matter
[8]). However, later in the report we will be calculating Tutte polynomials for
very large graphs, so we will instead use the Tutte polynomials relationship
to another polynomial called the chain polynomial (see the next section).
We are interested in the Tutte polynomial because of its relation to the
Jones polynomial for alternating links. The following theorem connecting
the Tutte polynomial to the Jones polynomial is given in [16] and [15]:
Theorem 3.1. Suppose a link L admits a connected, alternating oriented
link diagram D with a A-regions,

b B-regions and writhe w. Then the Jones
polynomial of L is given by the Tutte polynomial of the associated graph G
of the diagram D:
V
L
(t) = (1)
w
t

b a+3w
4
T[G]
_
t, t
1
_
.
17
3.3 The Chain Polynomial
3.3.1 Chain Graphs
As described in [16], we can obtain a chain graph from a graph (and then
use the chain polynomial of a chain graph to calculate the Tutte polynomial
of a graph). To obtain a chain graph from a graph:
1. Repeatedly remove vertices of degree two and join the two adjacent
vertices with a new edge.
2. Assign to each edge a label (e.g. a, b) with an associated positive
integer (e.g. n
a
= 2, n
b
= 5) that is equal to the number of edges the
new edge has replaced.
As noted above we will be obtaining standard graphs, G
L
, from the standard
diagram of a link L. We will denote by M
L
the chain graph obtained from
G
L
.
3.3.2 The Chain Polynomial
The chain polynomial, Ch[M](, {all labels}), is a polynomial that we as-
sociate with a chain graph M. It can be dened using ow polynomials of
subgraphs, or chromatic polynomials of graphs obtained by inserting chains
of edges [21]. (It was actually introduced to study chromatic polynomials of
such graphs [18].) Chain polynomials can be calculated using the following
relations, given in [18] and [21]:
1. If G is a graph with no edges, then
Ch[G] = 1 .
2. If an edge labelled a is a loop (i.e. its endpoints are the same vertex)
of G, then
Ch[G] = (a )Ch[Ga] .
If an edge labelled a is not a loop of G, then
Ch[G] = (a 1)Ch[Ga] +Ch[G/a] .
Where G a is the graph obtained from G by removing the edge labelled
a, and G/a is the graph obtained from G by contracting (i.e. removing and
identify the end vertices of) the edge labelled a.
18
We are interested in using chain polynomials to calculate Tutte polynomials
for families of graphs. The following theorem connecting the chain polyno-
mial to the Tutte polynomial is given with proof in [16] (we use = 1 q):
Theorem 3.2. In Ch[M], if we replace q by (1x)(1y), and replace each
label a by x
n
a
for every chain a, a polynomial

Ch[M] is obtained. Then
T[G](x, y) =
1
(x 1)
mn+k

Ch[M]
where m, n and k are the number of edges, vertices and components of the
chain graph M, respectively.
3.3.3 Properties
Lemma 3.3 [19]. If a is an edge of (a chain graph) M let H be the graph
obtained from M by deleting the edge a, and let K be the graph obtained
from M by contracting the edge a. Then
(i) Ch[H] is the coecient of a in Ch[M].
(ii) Ch[K] is obtained from Ch[M] by putting a = 1.
We can relate the Chain polynomial of a 2-vetex-connected chain graph to
the Chain polynomials of two chain graphs from which it is built. The
folowing lemma is given with proof (it is presented as a corollary of a more
general theorem) in [20].
Lemma 3.4. If G is a 2-vertex-connected graph as shown in Figure 3.2, and
G
1
and G
2
are formed from it as shown, then
Ch[G] = P
1
P
2

A
1
A
2
where
Ch[G
1
] = P
1
z +A
1
and Ch[G
2
] = P
2
z +A
2
.
We can generalise this to inating edges by a graph. By this we mean to
remove an edge from one graph, and replace it by another graph with a
similarly labelled edge removed. For example in Figure 3.2, we would be
inating the edge z in G
1
by G
2
, obtaining G.
Lemma 3.5. Let G
1
be a chain graph with n edges labelled z and let G
2
be a chain graph with one edge labelled z, i.e.
Ch[G
1
] =
n

i=0
c
i
z
i
and Ch[G
2
] = Pz +A .
19
z z
G
G
1
G
2
Figure 3.2: A graph G formed from two other graphs
If G is formed by inating all edges labelled z in G
1
by G
2
then
Ch[G] =
n

i=0
c
i
_

_
ni
P
i
.
Proof. See Appendix C, Section C.1.
Lemma 3.6. If G is formed by inating all edges in a sheaf chain graph by
a chain graph G
2
with Ch[G
2
] = Pz +A, then
Ch[G] =
1
1
__
P +
A

_
n
(P +A)
n
_
.
Proof. Substituting the coecients of z given in equation 3.3 into the equa-
tion from Lemma 3.5 obtains
Ch[G] =
n

i=0
(1)
n1
_
n
n i
__

ni

1
__

_
ni
P
i
=
1
1
__
P +
A

_
n
(P +A)
n
_
.
3.3.4 Useful Results
The following results for chain polynomials will be useful when we later
calculate Jones polynomials of families of links.
If M is a sheaf chain graph (i.e. a chain graph consisting of two vertices
joined by n parallel edges) with edges labelled a
1
, a
2
, . . . , a
n
, then [18]
Ch[M] =
1
1
_
n

i=1
(a
i
)
n

i=1
(a
i
1)
_
. (3.1)
20
So,
Ch
_

_
a b
_

_
=
1
1
_
(b ) (a )
n1
(b 1) (a 1)
n1
_
(3.2)
where there are n 1 edges labelled a. (This result is given in [16].)
And,
Ch
_

_
z
_

_
=
1
1
[(z )
n
(z 1)
n
]
=
n

i=0
z
i
(1)
n1
_
n
n i
__

ni

1
_
(3.3)
where there are n edges labelled z.
And,
Ch
_

_
a
b
_

_
=
1
1
[(b )
n
(a )
n
(b 1)
n
(a 1)
n
] (3.4)
where there are n edges labelled a and n edges labelled b.
21
Chapter 4
Torus Links
In this chapter, we will study a family of links called the torus links. Af-
ter establishing the denition and our notation, we will present the Jones
polynomials for torus knots and two types of torus links.
4.1 Denition and Notation
Torus links are knots and links that lie on the unkotted torus in S
3
[3].
We can dene a torus link by the number of times it wraps around a torus
in each direction: if it wraps around the meridian p times and around the
longitude q times then we call it the (p, q) torus link [3]. See Figure 4.1 for
a standard diagram of the (p, q) torus link. We will denote the (p, q) torus
link by T(p, q).
q
p
{
{
Figure 4.1: A standard diagram of the (p, q) torus link
Some properties of torus links:
1. T(p, q) is equivalent to T(q, p) [3].
22
2. If p and q are relatively prime (i.e. gcd(p, q) = 1) then T(p, q) is a
knot [3].
3. If p = 1 or q = 1 (the other arbitrary) then T(p, q) is the unkot [4].
4.2 Jones polynomial
4.2.1 Torus Knots
The (p, q) torus link is a knot (i.e. it has one component) when p and q are
relatively prime. In this case the Jones polynomial is [2]
V
T(p,q)
=
t
(p1)(q1)
2
1 t
2
_
1 t
p+1
t
q+1
+t
p+q
_
. (4.1)
4.2.2 (p, q) Torus Links with p = 2, 3
Where p and q are not relatively prime, the (p, q) torus knot is a link of 2
or more components.
If p = 2 then, for any q 1
V
T(2,q)
= (1)
q+1
_
t
q1
2
+
t
q+1
2
1 +t
((t)
q
+t)
_
. (4.2)
The proof of this is given in Appendix B, Section B.2.
The Jones polynomial for a (p,q) torus knot/link with p = 3 is given in [10]
as:
V
T(3,q)
=
_
t
q1
+t
q+1
+ 2t
2q
if q 0 (mod 3),
t
q1
+t
q+1
t
2q
otherwise.
(4.3)
23
Chapter 5
Pretzel Links
In this chapter, we will study a family of links called the pretzel links.
After establishing the denition and our notation, we will obtain general
expressions for the Jones polynomials of three particular families of pretzel
links.
5.1 Denition and Notation
{
p
1
p
2
p
n
{ {
Figure 5.1: General pretzel link Figure 5.2: (2,3,2,2,5) pretzel link
A pretzel link is formed by joining tangles in a cyclic fashion, where each
tangle consists of two vertical strings twisted together. We describe a pretzel
link by the n-tuple of the number of half-twists (i.e. crossings) in each tangle
(the sign given by Figure 5.3). The standard diagram of the (p
1
, p
2
, ..., p
n
)
pretzel link is given in Figure 5.1 and an example given in Figure 5.2.
Continuing the notation of [16], we will denote the (k
1
, k
2
, ..., k
n
) pretzel link
by P(k
1
, k
2
, ..., k
n
). Also denote the (
n
..
k
1
, ..., k
1
) pretzel link by P (k
1
(n)).
Similarly, denote the (
n
..
k
1
, ..., k
1
,
m
..
k
2
, ..., k
2
) pretzel link by P(k
1
(n), k
2
(m)).
24
+ve -ve
Figure 5.3: Sign of half-twists in a pretzel link
5.2 Jones polynomial
5.2.1 (k(n)) Pretzel Links
In [16] the Jones polynomial (and its zeros) of P(3(n)) are studied. Here,
we will use the same methods to study the more general case of P(k(n)) for
k > 0.
{
k k k
{ {
a a a
Figure 5.4: Diagram of P(k(n)) with associated graph, G
P(k(n))
, and
associated chain graph, M
P(k(n))
(where there are n edges labelled a
and n
a
= k).
The pretzel link containing n tangles all with k > 0 half-twists, P(k(n)),
admits an associated chain graph that is a sheaf graph (i.e. a graph con-
sisting of two vertices joined by n edges) with every edge having the same
label (see Figure 5.4). Then by equation 3.3 and Theorem 3.2 we have the
Tutte polynomial for the associated graph of P(k(n)):
T[G
P(k(n))
] =
1
(x 1)
n
(y 1)
_
(x
k
x +xy y)
n
+ (x
k
1)
n
(xy x y)
_
.
By inspection of the link diagram we can see that a = 2 + (k 1)n and

b = n. Then by Theorem 3.1 we have


V
P(k(n))
=
(1)
kn
t
2nkn2+3w
4
(t 1)
n
(t
1
1)

__
(t)
k
+t + 1 +t
1
_
n
+
_
t + 1 +t
1
_
_
(t)
k
1
_
n
_
.
(5.1)
25
Where w depends on the signs of k and n, and the orientation given to the
components of the link.
5.2.2 (k(1), l(n 1)) Pretzel Links
In [16] the Jones polynomial (and its zeros) of P(k(1), 1(n 1)) and
P (k(1), 2(n 1)) are studied. Here, we will use the same methods to study
the more general case of P(k(1), l(n 1)) for l > 0 and k > 0.
{
k l l
{ {
b a a
Figure 5.5: Diagram of P(k(1), l(n 1)) with associated graph,
G
P(k(1),l(n1))
, and associated chain graph, M
P(k(1),l(n1))
(where there
are n 1 edges labelled a, n
b
= k and n
a
= l).
The pretzel link containing n tangles one with k half-twists and the remain-
ing with l half-twists, P(k(1), l(n 1)), admits an associated chain graph
that is a sheaf graph with all but one edge having the same label (see Figure
5.5). Then by equation 3.2 and Theorem 3.2 we have the Tutte polynomial
for the associated graph of P(k(1), l(n 1)):
T[G
P(k(1),l(n1))
] =
1
(x 1)
n
(y 1)
_
(x
l
x +xy y)
n1
(x
k
x +xy y)
+(x
l
1)
n1
(x
k
1)(xy x y)
_
.
By inspection of the link diagram we can see that a = l + nk k n + 2
and

b = n. Then by Theorem 3.1 we have
V
P(k(1),l(n1))
=
(1)
w
t
2nlnk+k2+3w
4
(t 1)
n
(1 t
1
)

_
_
(t)
l
+t + 1 +t
1
_
n1
_
(t)
k
+t + 1 +t
1
_
+
_
(t)
l
1
_
n1
_
(t)
k
1
_
_
t + 1 +t
1
_
_
.
(5.2)
Where w depends on the signs of l, k and n, and the orientation given to
the components of the link.
26
5.2.3 (k(n), l(n)) Pretzel Links
{
k l
l
{
{
b a
k
{
Figure 5.6: Diagram of P(k(n), l(n)) with associated graph, G
P(k(n),l(n))
,
and associated chain graph, M
P(k(n),l(n))
(where there are n edges la-
belled a, n edges labelled b, n
b
= k and n
a
= l).
The pretzel link containing 2n tangles half with k half-twists and half with
l half-twists, P(k(n), l(n)), admits an associated chain graph that is a sheaf
graph with half of the edges having the same label (see Figure 5.6). Then
by equation 3.4 and Theorem 3.2 we have the Tutte polynomial for the
associated graph of P(k(n), l(n)):
T[G
P(k(n),l(n))
] =
1
(x 1)
n
(y 1)
__
x
l
x +xy y
_
n
_
x
k
x +xy y
_
n
+
_
x
l
1
_
n
_
x
k
1
_
n
(xy x y)
_
.
By inspection of the link diagram we can see that a = ln +kn 2n +2 and

b = 2n. Then by Theorem 3.1 we have


V
P(k(n),l(n))
=
(1)
w
t
4nlnkn2+3w
4
(t 1)
n
(1 t
1
)

__
(t)
l
+t + 1 +t
1
_
n
_
(t)
k
+t + 1 +t
1
_
n
+
_
(t)
l
1
_
n
_
(t)
k
1
_
n _
t + 1 +t
1
_
_
.
(5.3)
Where w depends on the signs of l, k and n, and the orientation given to
the components of the link.
27
Chapter 6
Montesinos Links
In this chapter, we will study some of the family of links called the Mon-
tesinos links. After establishing the denition and our notation, we will
obtain general expressions for the Jones polynomials of two particular fam-
ilies of Montesinos links.
6.1 Denition and Notation
6.1.1 Rational Tangles
Figure 6.1: An example (rational)
tangle
Figure 6.2: An example (rational)
tangle
A tangle is a link with four free ends which touch a bounding sphere [3],
two examples are show in Figures 6.1 and 6.2. If we imagine the four free
ends of a tangle to be restricted to the sphere, then a tangle that we can
deform into two separate strands by only moving the free ends on the
sphere is called a rational tangle [22]. (The tangle in Figure 6.2 is a rational
tangle.) All rational tangles are equivalent to a basic tangle that can be
28
formed in the following way [22]:
1. Start with a horizontal (or vertical) tangle
2. Add a vertical (or horizontal) tangle to the bottom (or right)
3. Add a horizontal (or vertical) tangle to the right (or bottom)
4. Repeat these steps a nite number of times
(Where a horizontal (or vertical) tangle is formed by starting with two par-
allel horizontal (or vertical) strands and twisting the endpoints. Each hori-
zontal (or vertical) tangle has an associated integer equal to the number of
half-twists (or crossings) it contains, where sign depends as shown in Figure
6.3.) When we start with a horizontal (or vertical) tangle, then the result
is called a basic horizontal (or vertical) tangle [22].
A horizontal tangle, with associated integer +4.
A horizontal tangle, with associated integer -3.
A vertical tangle,
with associated integer +3.
A vertical tangle,
with associated integer -2.
Figure 6.3: Horizontal and vertical tangles
In this report we will use two particular rational tangles, shown in Figures
6.4 and 6.5. We will denote the rational tangle shown in Figure 6.4 by {k, l}
and in Figure 6.5 by {m, k, l}.
6.1.2 Montesinos Links
The Montesinos links are formed in a similar way to the pretzel links, but
each tangle consists of a rational tangle. The standard diagram for the
(classic) Montesinos link is given in Figure 6.6, where each box (T
1
, . . . , T
n
)
contains a rational tangle.
29
k
{
l
{
Figure 6.4: {k, l} rational tangle
l
{
k
{
m
{
Figure 6.5: {m, k, l} rational tan-
gle
T
1
T
2
T
n
Figure 6.6: Standard diagram for a (classic) Montesinos link
In this report we will be using Montesinos links where all of the n tangles
are the same. We will call the Montesinos link where every tangle is the
{k, l} tangle (Figure 6.4), the ({k, l}(n)) Montesinos link and denote it by
M({k, l}(n)). Similarly, where every tangle is the {m, k, l} tangle (Figure
6.5), it is the ({m, k, l}(n)) Montesinos link and denote it by M({m, k, l}(n)).
6.2 Jones polynomial
6.2.1 ({k, l}(n)) Montesinos Links
In the family of links given by M({k, l}(n)) each tangle is as shown in
Figure 6.4. The associated graph of M({k, l}(n)) consists of n copies of
the subgraph shown in Figure 6.7, with all vertices labelled 1 identied
and all vertices labelled 2 identied. Then the associated chain graph of
M({k, l}(n)) consists of n copies of the subgraph shown in Figure 6.8, with
all vertices labelled 1 identied and all vertices labelled 2 identied.
To nd a general expression for the Jones polynomial of these links, rst
we obtain the chain polynomial of a graph from which we can build the
30
1
2
Figure 6.7: Subgraph of
G
M({k,l}(n))
a
b
1
2
Figure 6.8: Subgraph of
M
M({k,l}(n))
(n
a
= k, n
b
= 1,
l edges labelled b)
associated chain graph of M({k, l}(n)):
Ch
_

_
a
z
b
_

_
=
za
1
_
(b )
l
(b 1)
l
_
+
1
1
_
(b )
l
+ (b 1)
l
_
.
Where there are l edges labelled b. (The calculation of this is given in
Appendix C, Section C.2.) Then by Lemma 3.6 we obtain:
Ch[M
M({k,l}(n))
] =
1
1
_
(b 1)
l
(a 1) + (b )
l
(1 a)
1
_
n

1
1
_
(b 1)
l
(a ) + (b )
l
( a)
1
_
n
Setting n
a
= k and n
b
= 1, by Theorem 3.2 we have the Tutte polynomial
for the associated graph of M({k, l}(n)):
T[G
M({k,l}(n))
] =
(1)
ln+n
(1 x)
ln+n
(1 y)
n+1

__
(x 1)
l
(x
k
(x +y xy) 1) (xy y)
l
(x
k
1)
_
n
(x +y xy)

_
(x 1)
l
(x
k
1)(x +y xy) (xy y)
l
(x
k
(x +y xy))
_
n
_
By inspection of the link diagram we can see that a = kn + 2 and

b = nl.
Then by Theorem 3.1 we have
V
M({k,l}(n))
=
(1)
w+ln+1
t
n(lk)+3w
4
(t + 1)
(2+l)n+1

__
(t 1)
l
(A+t) (1 +t
1
)
l
(B +t)
_
n
(t
2
+t + 1)
+t
_
(t 1)
l
(At
2
t 1) (1 +t
1
)
l
(B t
2
t 1)
_
n
_
(6.1)
31
with A = (t)
k+2
(t)
k+1
+ (t)
k
and B = (t)
k+1
. Where w depends
on the signs of l, k and n, and the orientation given to the components of
the link.
6.2.2 ({m, k, l}(n)) Montesinos Links
In the family of links given by M({m, k, l}(n)) each tangle is as shown in
Figure 6.5. The associated graph of M({m, k, l}(n)) consists of n copies
of the subgraph shown in Figure 6.9, with all vertices labelled 1 identied
and all vertices labelled 2 identied. Then the associated chain graph of
M({m, k, l}(n)) consists of n copies of the subgraph shown in Figure 6.10,
with all vertices labelled 1 identied and all vertices labelled 2 identied.
1
2
Figure 6.9: Subgraph of
G
M({m,k,l}(n))
a
b
1
2
c
Figure 6.10: Subgraph of
M
M({m,k,l}(n))
(n
a
= k, n
b
= 1,
n
c
= 1, l edges labelled b, m edges
labelled c)
To nd a general expression for the Jones polynomial of these links, rst
we obtain the chain polynomial of a graph from which we can build the
associated chain graph of M({m, k, l}(n)):
Ch
_
a
z
b
c
_
=
z
(1 )
2
_
a
_
(b )
l
(b 1)
l
_
((c )
m
(c )
m
)
+
_
(b )
l
+(b 1)
l
_
((c )
m
(c 1)
m
)
_
+
1
(1 )
2
_
(b )
l
+(b 1)
l
_
((c )
m
(c 1)
m
)

1
(1 )
2
_
(1 +)((b )
l
+(b 1)
l
)
a((b )
l
(b 1)
l
)
_
((c )
m
(c 1)
m
)
Where there are l edges labelled b and m edges labelled c. (The calculation
32
of this is given in Appendix C, Section C.2.) Then by Lemma 3.6 we obtain:
Ch[M
M({m,k,l}(n))
] =
1
1
_
_
(b 1)
l
(a 1) + (b )
l
(1 a)
_
(c 1)
m
1
_
n

1
1
_
_
(b 1)
l
(a ) + (b )
l
( a)
_
(c )
m
1
_
n
Setting n
a
= k, n
b
= 1 and n
c
= 1, by Theorem 3.2 we have the Tutte
polynomial for the associated graph of M({m, k, l}(n)):
T[G
M({m,k,l}(n))
] =
(1)
(l+m+1)n
(1 x)
(l+m+1)n
(1 y)
n+1

__
((x 1)
l
(x
k
(x +y xy) 1) (xy y)
l
(x
k
1))(x 1)
m
_
n
(x +y xy)
_
((x 1)
l
(x
k
1)(x +y xy) (xy y)
l
(x
k
(x +y xy)))(xy y)
m
_
n
_
By inspection of the link diagram we can see that a = kn+2 and

b = (l+m)n.
Then by Theorem 3.1 we have
V
M({m,k,l}(n))
=
(1)
(l+m)n++1
t
(l+mk)n2+3w
4
t
mn
(t + 1)
(l+2)n+1

__
((t 1)
l
(A+t) (1 +t
1
)
l
(B +t))(t)
m
_
n
(t
2
+t + 1)
+t
_
(t 1)
l
(At
2
t 1) (1 +t
1
)
l
(B t
2
t 1)
_
n
_
(6.2)
with A = (t)
k+2
(t)
k+1
+ (t)
k
and B = (t)
k+1
. Where w depends
on the signs of l, k and n, and the orientation given to the components of
the link.
33
Chapter 7
Zeros of the Jones
Polynomial
As discussed in Chapter 2, the link between the Jones polynomial and the
Potts model partition function (and the importance of zeros of partition
functions to thermodynamic properties) motivates us to study zeros of Jones
polynomials. In particular we wish to nd the accumulation sets of zeros for
families of links as the number of crossings tends to ininity. In this chapter
we will use the general expressions for Jones polynomials found in Chapters
4, 5 and 6 to nd some of these accumulation sets.
Whilst the Jones polynomial is not a true polynomial it can always be ex-
pressed as a polynomial in t multiplied by some factor [5], so we are justied
in nding zeros [18]. We will only consider the non-zero zeros of Jones poly-
nomials (so we can assume t = 0 in our calculations) and our plots will
also not include any t = 0 solutions to numerical examples. Also, since
the Jones polyomial of a knot does not depend on orientation [2] and the
Jones polynomials of links only dierer by a factor of t
a
(for some integer a)
for dierent orientations [16], we do not need to consider orientation when
nding zeros in this chapter.
7.1 General Knots and Links
The Jones polynomial has the property [6]
V
L
(e
2i
3
) = (1)
n1
34
where L is a link of n components. This suggests that generally V
L
(t) may
have zeros near t = e

2i
3
[15]. For knots (where n = 1) the Jones polynomial
has the property [6]
V
K
(t) = 1 (1 t)
_
1 t
3
_
W
K
(t)
where W
K
(t) is a Laurent polnomial. So V
K
(t) has zeros approaching t =
e

2i
3
as the (minimum) number of crossings tends to innity [14] [15]. This
can be seen in the plots of zeros for particular knots and links given in this
chapter.
7.2 Torus Links
7.2.1 Torus Knots
It is shown in [14] that the zeros of T(p, q) are distributed uniformly on
the unit circle |t| = 1 as q for any (xed) p. An example with p = 2
is given in Figure 7.1 and with p = 5 in Figure 7.2.
7.2.2 (2, q) Torus Links
Equating equation 4.2 to zero and rearranging (and we assume t = 0), we
obtain
(t)
q
=
t
2
+t + 1
t
.
Taking the q
th
root, gives us
t

t
2
+t + 1
t

1
q

, = 1, 2, ..., n,
with = e
2i
n
. Now as q we have t

n
. So the zeros of V
T(2,q)
are distributed on the unit circle as q . A numerical example (with
q = 40) and the unit circle are shown in Figure 7.3.
7.2.3 (3, q) Torus Links
T(3, q) is a link of more than one component when q 0 (mod 3). Equating
the relevant part of equation 4.3 to zero and rearranging (and we assume
t = 0), we obtain
t
q
=
t t
1
2
.
35
Taking the q
th
root, gives us
t

t +t
1
2

1
q

, = 1, 2, ..., n,
with = e
2i
n
. Now as q we have t

n
. So the zeros of a link
V
T(3,q)
are distributed on the unit circle as q . A numerical example
(with q = 36) and the unit circle are shown in Figure 7.4.
Figure 7.1: Zeros of V
T(2,29)
and of
V
T(p,q)
as q for any xed p
Figure 7.2: Zeros of V
T(5,32)
and of
V
T(p,q)
as q for any xed p
Figure 7.3: Zeros of V
T(2,q)
with
q = 40 as q
Figure 7.4: Zeros of V
T(3,q)
with
q = 36 and as q
36
7.3 Pretzel Links
7.3.1 (k(n)) Pretzel Links
Equating the numerator of equation 5.1 to zero and rearranging (and we
assume t = 0), we obtain
_
(t)
k
+t + 1 +t
1
(t)
k
1
_
n
= (t + 1 +t
1
) .
Taking the n
th
root and n , we obtain the set
|(t)
k
+t + 1 +t
1
| = |(t)
k
1|
on which the zeros of V
P(k(n))
are distributed as n . An example of the
curves in the complex plane given by the set with k = 4 is given in Figure
7.5 and with k = 5 in Figure 7.6.
Figure 7.5: Zeros of V
P(4(n))
with
n = 20 and as n
Figure 7.6: Zeros of V
P(5(n))
with
n = 20 and as n
7.3.2 (k(1), l(n 1)) Pretzel Links
Equating the numerator of equation 5.2 to zero and rearranging (and we
assume t = 0), we obtain
_
(t)
l
+t + 1 +t
1
(t)
l
1
_
n1
=
((t)
k
1)(t + 1 +t
1
)
(t)
k
+t + 1 +t
1
.
37
Taking the (n 1)
th
root and n , we obtain the set
|(t)
l
+t + 1 +t
1
| = |(t)
l
1|
on which the zeros of V
P(k(1),l(n1))
are distributed as n . This result
is identical to that for (l(n)) pretzel links, found in the previous section.
7.3.3 (k(n), l(n)) Pretzel Links
Equating the numerator of equation 5.3 to zero and rearranging (and we
assume t = 0), we obtain
_
((t)
l
+t + 1 +t
1
)((t)
k
+t + 1 +t
1
)
((t)
l
1)((t)
k
1)
_
n
= (t + 1 +t
1
) .
Taking the n
th
root and n , we obtain the set
|[(t)
k
+t + 1 +t
1
][(t)
l
+t + 1 +t
1
]| = |[(t)
k
1)][(t)
l
1]|
on which the zeros of V
P(k(n),l(n))
are distributed as n . An example
with k = 2 and l = 4 is given in Figure 7.7 and with k = 4 and l = 5 in
Figure 7.8.
Figure 7.7: Zeros of V
P(2(n),4(n))
with n = 10 and as n
Figure 7.8: Zeros of V
P(4(n),5(n))
with n = 10 and as n
38
7.4 Montesinos Links
7.4.1 ({k, l}(n)) Montesinos Links
Equating the numerator of equation 6.1 to zero and rearranging (and we
assume t = 0), we obtain
_
(t 1)
l
(A+t) (1 +t
1
)(B +t)
(t 1)
l
(At
2
t 1) (1 +t
1
)(B t
2
t 1)
_
n
=
t
t
2
+t + 1
.
with A = (t)
k+2
(t)
k+1
+(t)
k
and B = (t)
k+1
. Taking the n
th
root
and n , we obtain the set
|(t1)
l
(A+t)(1+t
1
)(B+t)| = |(t1)
l
(At
2
t1)(1+t
1
)(Bt
2
t1)| .
Dividing by (1 +t
1
) we obtain the set
|(t)
l
[A+t] [B +t]| = |(t)
l
[At
2
t 1] [B t
2
t 1]|
on which the zeros of M({k, l}(n)) are distributed as n . An example
with l = 2 and k = 3 is given in Figure 7.9 and with l = 4 and k = 5 in
Figure 7.10.
Figure 7.9: Zeros of V
M({3,2}(n))
with n = 15 and as n
Figure 7.10: Zeros of V
M({5,4}(n))
with n = 10 and as n
39
7.4.2 ({m, k, l}(n)) Montesinos Links
Equating the numerator of equation 6.2 to zero and rearranging (and we
assume t = 0), we obtain
_
((t 1)
l
(A+t) (1 +t
1
)(B +t))(t)
m
(t 1)
l
(At
2
t 1) (1 +t
1
)(B t
2
t 1)
_
n
=
t
t
2
+t + 1
.
with A = (t)
k+2
(t)
k+1
+(t)
k
and B = (t)
k+1
. Taking the n
th
root
and n , we obtain the set
|((t 1)
l
(A+t) (1 +t
1
)(B +t))(t)
m
|
= |(t 1)
l
(At
2
t 1) (1 +t
1
)(B t
2
t 1)| .
Dividing by (1 +t
1
) we obtain the set
|(t)
m
((t)
l
[A+t] [B +t])| = |(t)
l
[At
2
t 1] [B t
2
t 1]|
on which the zeros of M({m, k, l}(n)) are distributed as n . An example
with l = 2, k = 3 and m = 2 is given in Figure 7.11 and with l = 2, k = 5
and m = 2 in Figure 7.12.
Figure 7.11: Zeros of V
M({2,3,2}(n))
with n = 10 and as n
Figure 7.12: Zeros of V
M({2,5,2}(n))
with n = 10 and as n
40
Chapter 8
Conclusion
After introducting the basic ideas of knot theory, we looked at the relation
between the Jones polynomial and the partition function of the Potts mdel.
This motivated us to continue the investigation of zeros of the Jones Polyno-
mial. In particular, we found general expressions for the Jones polynomial
for families of links, so that we could nd accumulation sets of zeros of the
Jones polynomial as the number of crossings in the links tended to innity.
The work of Wu and Wang [14], Chang and Shrock [15] and Jin and Zhang
[16], [17] and [18], and this report concentrate on important families of
links: the torus links, the pretzel links and the Monesinos links. There
are still many links in these families whose Jones polynomials have not
been studied in this way, and future investigation into these is possible. In
particular, Montesinos knots containing more complex rational tangles and
even dierent rational tangles could be investigated.
41
Bibliography
[1] C.M. Livingston, Knot Theory, Math. Assoc. America, (1993).
[2] W.B.R. Lickorish, An Introduction to Knot Theory, Springer, (1997).
[3] C.A. Adams, The Knot Book : An Elementary Introduction to
the Mathematical Theory of Knots, American Mathematical Soci-
ety, (2004).
[4] V. Manturov, Knot Theory, Chapman & Hall/CRC, (2004).
[5] V.F.R. Jones, A polynomial invariant for knots via Von Neumann
algebras,
Bulletin of the American Mathematical Society, 12 (1987) 103-111.
[6] V.F.R. Jones, Hecke algebra representations of braid groups and link
polynomials,
Annals of Mathematics, 126 (1987) 335-388.
[7] L.H. Kauman, State models and the Jones polynomial,
Topology, 26 (1987) 395-407.
[8] L. Beaudina, J. Ellis-Monaghanb, G. Pangbornc and R. Shrockd, A
little statistical mechanics for the graph theorist,
Discrete Mathematics, (2010).
[9] R.A. Landvoy, The Jones polynomial of pretzel knots and links,
Topology and its Applications, 83 (1998) 135-147.
[10] Yoshihiro Hirata, Hisanori Naka and Yaichi Shinohara, On the Jones
polynomial for the torus links T
p,q
with p = 3,4,5,
Kwansei Gakuin University natural sciences review, 5 (2000) 1-16.
[11] Yaichi Shinohara and Kazunari Uetani, On the Jones Polynomial of
Pretzel Links,
Kwansei Gakuin University natural sciences review, 8 (2003) 1-16.
42
[12] C.N. Yang and T.D. Lee, Statistical Theory of Equations of State
and Phase Transitions. I. Theory of Condensation,
Physical Review, 87 (1952) 404-409.
[13] C.N. Yang and T.D. Lee, Statistical Theory of Equations of State
and Phase Transitions. II. Lattice Gas and Ising Model,
Physical Review, 87 (1952) 410-419.
[14] F.Y. Wu and J. Wang, Zeroes of the Jones polynomial,
Physica A: Statistical Mechanics and its Applications, 296 (2001)
483-494.
[15] S.-C. Chang and R. Shrock, Zeros of Jones polynomials for families
of knots and links,
Physica A: Statistical Mechanics and its Applications, 301 (2001)
196-218.
[16] X. Jin and F. Zhang, Zeros of the Jones polynomials for families of
pretzel links,
Physica A: Statistical Mechanics and its Applications, 328 (2003)
391-408.
[17] X. Jin and F. Zhang, Jones polynomials and their zeros for a family
of links,
Physica A: Statistical and Theoretical Physics, 333 (2004) 183-196.
[18] X. Jin, F. Zhang, F. Dong and E.G. Tay, On zeros of the Jones
polynomial,
http://59.77.1.114/Upload/le/bigFile/jxa/papers/zerosJones.pdf,
(2009).
[19] Ronald C. Read and Earl Glen Whitehead, Jr., Chromatic polyno-
mials of homeomorphism classes of graphs,
Discrete Mathematics, 204 (1999) 337-356.
[20] Ronald C. Read, Chain polynomials of graphs,
Discrete Mathematics, 265 (2003) 213-235.
[21] L. Traldi, Chain polynomials and Tutte polynomials,
Discrete Mathematics, 248 (2002) 279-282.
[22] J R. Goldman and L.H. Kauman, Rational Tangles,
Advances in Applied Mathematics, 18 (1997) 300-332.
[23] KnotPlot 1.0, www.knotplot.com
[24] Adobe Illustrator 10
[25] Maple 12
43
Appendix A
A Knot Table
Figure A.1: A knot table
Figure A.1 is reproduced from http://en.wikipedia.org/wiki/File:Knot_
table.svg (Author: Jkasd, 2008).
Figure A.1 is a table of the prime knots (not including mirror images) with
up to 7 crossings. The knots are labelled with the number of crossings, and
the subscript just denotes an order of the knots.
44
Appendix B
Jones Polynomial Results
B.1 Jones polynomials of T(2,1) and T(2,2)
Figure B.1: T(2, 2) Figure B.2: T(2, 1)
By Lemma 2.2 we have:
=A
3
= A
3
=A
3
= A
3
Then using the denition of the Kauman bracket:
=A +A
1

=A
_
A
3
_
+A
1
_
A
3
_
=A
4
A
4
The writhe of the oriented link is clear from the diagram:
w
_ _
= 2
45
So by the denition of the Jones polynomial via the Kauman bracket:
V
T(2,2)
= V
_ _
=
_
t
1
4
_
32

_
t
1
4
_
4

_
t
1
4
_
4
__
=t
3
2
_
t t
1
_
=t
1
2
t
5
2
(B.1)
By a Type I Reidemeister move on the diagram shown on Figure B.2 we can
easily see that T(2, 1) is equivalent to the unkot, so:
V
T(2,1)
= 1 (B.2)
B.2 Proof of equation 4.2
Proof. From equations B.1 and B.2 we can see that
V
T(2,2)
=t
1
2
t
5
2
= (1)
2+1
_
t
21
2
+
t
2+1
2
1 +t
((t)
2
+t)
_
and
V
T(2,1)
=1 = (1)
1+1
_
t
11
2
+
t
1+1
2
1 +t
((t)
1
+t)
_
.
Given the skein relation of the Jones polynomial, we have
(t
1
2
t

1
2
)V
T(2,q1)
= t
1
V
T(2,q)
tV
T(2,q2)
. (B.3)
Assume we have q 1 and q 2 such that
V
T(2,q1)
=(1)
(q1)+1
_
t
(q1)1
2
+
t
(q1)+1
2
1 +t
((t)
(q1)
+t)
_
and
V
T(2,q2)
=(1)
(q2)+1
_
t
(q2)1
2
+
t
(q2)+1
2
1 +t
((t)
(q2)
+t)
_
.
(B.4)
Then substituting B.4 into B.3 obtains
(t
1
2
t

1
2
)(1)
(q1)+1
_
t
(q1)1
2
+
t
(q1)+1
2
1 +t
((t)
(q1)
+t)
_
=t
1
V
T(2,q)
t(1)
(q2)+1
_
t
(q2)1
2
+
t
(q2)+1
2
1 +t
((t)
(q2)
+t)
_
46
V
T(2,q)
=t(1)
q
_
t
q1
2
+
t
q+1
2
1 +t
((t)
q1
+t)
_
t(1)
q
_
t
q3
2
+
t
q1
2
1 +t
((t)
q1
+t)
_
+t(1)
q1
_
t
q1
2
+
t
q+1
2
1 +t
((t)
q2
+t)
_
=(1)
q
_
t
q+1
2
+
t
q+3
2
1 +t
((t)
q1
+t)
_
(1)
q
_
t
q1
2
+
t
q+1
2
1 +t
((t)
q1
+t)
_
(1)
q
_
t
q+1
2
+
t
q+3
2
1 +t
((t)
q2
+t)
_
=(1)
q
_
t
q1
2
+
1
1 +t
_
(t)
q1
t
q+3
2
+t
q+5
2
(t)
q1
t
q+1
2
t
q+3
2
(t)
q2
t
q+3
2
t
q+5
2
_
_
=(1)
q
_
t
q1
2
+
t
q+3
2
1 +t
_
(t)
q1
1 (t)
q2
+ (t)
q2
_
_
=(1)
q
_
t
q1
2

t
q+1
2
1 +t
((t)
q
+t)
_
=(1)
q+1
_
t
q1
2
+
t
q+1
2
1 +t
((t)
q
+t)
_
So, by induction,
V
T(2,q)
= (1)
q+1
_
t
q1
2
+
t
q+1
2
1 +t
((t)
q
+t)
_
for all q 1.
47
Appendix C
Chain Polynomial Results
This appendix will include calculations on Chain polynomials that are not
included in the main text.
C.1 Proof of Lemma 3.5
Proof. Let G
(k)
be the graph formed by inating k edges labelled z in G
1
by G
2
, so G
(n)
is G. We have
Ch[G
1
] =
n

i=0
c
i
z
i
=
_
n

i=1
c
i
z
i1
_
z +c
0
.
So by Lemma 3.4
Ch[G
(1)
] =
_
n

i=1
c
i
z
i1
_
P
A

c
0
=
_
n

i=2
c
i
z
i2
_
Pz +Pc
1

c
0
.
48
Then repeatedly applying Lemma 3.4 we obtain
Ch[G
(2)
] =
_
n

i=2
c
i
z
i2
_
P
2

_
Pc
1

c
0
_
=
_
n

i=3
c
i
z
i3
_
P
2
z +P
2
c
2

Pc
1
+
_

_
2
c
0
Ch[G
(3)
] =
_
n

i=3
c
i
z
i3
_
P
3

_
P
2
c
2

Pc
1
+
_

_
2
c
0
_
=
_
n

i=4
c
i
z
i4
_
P
3
z +P
3
c
3

P
2
c
2
+
_

_
2
Pc
1
+
_

_
3
c
0
.
.
.
Ch[G
(n
)] =P
n
c
n
+
_

_
P
n1
c
n1
+
_

_
2
P
n2
c
n2
+. . . +
_

_
n1
Pc
1
+
_

_
n
c
0
=
n

i=0
c
i
_

_
ni
P
i
.
C.2 Calculation of Some Chain Polynomials
b
a
c
d
e
f
Figure C.1: K
4
graph
a
c
d
Figure C.2: K
3
graph
The chain polynomial of the complete graph with 4 vertices, K
4
(labelled as
shown in Figure C.1), is given in [19] as
Ch[K
4
] =abcdef (adf +abc +bef +cde) (ae +bd +cf)
+ ( +
2
)(a +b +c +d +e +f) (2 + 3
2
+
3
) .
49
The complete graph with 3 vertices, K
3
(labelled as shown in Figure C.2),
can be formed from K
4
by removing edges b and e and contracting edge f.
So by Lemma 3.3 we obtain
Ch[K
3
] = acd . (C.1)
By equation 3.2 we also have
Ch
_

_
b c
_

_
=
1
1
_
(c )(b )
l
(c 1)(b 1)
l
_
=
c
1
_
(b )
l
(b 1)
l
_
+
1
1
_
(b )
l
+(b 1)
l
_
(C.2)
where there are l edges labelled b.
Applying Lemma 3.4 to equations C.1 and C.2 (and relabelling d as z) we
obtain
Ch
_

_
a
z
b
_

_
=
za
1
_
(b )
l
(b 1)
l
_

()
1
1
_
(b )
l
+(b 1)
l
_
=
za
1
_
(b )
l
(b 1)
l
_
+
1
1
_
(b )
l
+(b 1)
l
_
(C.3)
where there are l edges labelled b.
By equation 3.1 we have
Ch
_
_ c z z
_
_
=
1
1
[(c )
m
(z )
2
(c 1)
m
(z 1)
2
] (C.4)
where there are m edges labelled c. Aplying Lemma 3.4 to equations C.3
50
and C.4 we obtain
Ch
_

_
a
z
b
c
_

_
=
z
(1 )
2
_
a
_
(b )
l
(b 1)
l
_
((c )
m
(c )
m
)
+
_
(b )
l
+(b 1)
l
_
((c )
m
(c 1)
m
)
_
+
1
(1 )
2
_
(b )
l
+(b 1)
l
_
((c )
m
(c 1)
m
)

1
(1 )
2
_
(1 +)((b )
l
+(b 1)
l
)
a((b )
l
(b 1)
l
)
_
((c )
m
(c 1)
m
)
where there are l edges labelled b and m edges labelled c.
51

Você também pode gostar