Computing the colouring invariant of a knot

The mathematical study of knots involves many different mathematical techniques and has various facets. While in everyday usage the term "knot" usually refers to a configuration of an open ended string, from the mathematical point of view (the topological point of view to be precise) these are not knots at all since they can easily be untied and deformed into a straight piece of string. For mathematical purposes, a knot is considered to be a continuous closed curve in 3-dimensional space. Many of the keys questions concerning mathematical knots are related to the problem of distinguishing two essentially different knots.

Two knots are considered to be the same if they can be transformed one into the other in 3 dimensional space without resorting to scissors. For example, most people would agree that the following pairs of knots are equivalent.

In each of the above diagrams the knots are represented by projections onto the 2 dimensional plane (of the video monitor or paper if you print this out). In 3 dimensional space the string never intersects itself but in a projection this may appear to be the case. If the string is represented as having a non-zero thickness then this will not usually create any problem in interpreting the diagram. However, knots are usually represented as lines without thickness (recall the mathematical definition of a knot a curve in 3-space), If care is not taken to deal with crossings, an ambiguous represention may result. This is dealt with by representing a crossing so that the part of the string passing underneath the other part is pictured as broken even though, in reality, it is not. For example, the last two knots would be represented as follows

If a knot represented in this way has k crossings then the string will appear to be broken into k distinct arcs as well. It is important to note that the same knot can be projected onto the plane in various different ways. However, by studying the provided examples more closely, one quickly comes to the conclusion that a knot is not essentially changed by performing certain transformations to it. For example, suppose that
this is transformed to this and then this
By repeating the mirror image of this sequence of moves one get the following
from this one gets this and then this
This is known both of these are known as Reidemeister moves of type I. One sees that the first pair of knots above can be transformed into each other by a move of this type. Moreover, it should be clear that moves of this type do not change the basic nature of a knot. There are two other type of Reidemeister moves. The second type is shown i nthe following sequence of moves. 

A sequence of moves like this is known as Reidemeister move of type II. A Reidemeister move of type II will transform the two two knots of the second pair above. There are also Reidemeister moves of type III. These are shown below.

A Reidemeister move of type III will transform the last pair of knots into each other. Since it has already been noted that the Reidemeister moves of type I, II or III will not essentially change the nature of a knot, the same must be tru for any sequence of Reidemeister moves. It is not a little surprising that these three very simple type of moves are all that is necessary to exhibit that two knots are the same. Reidemeister's Theorem says that two knots are equivalent if and only if one can be transformed into the other by applying a sequence of Reidemeister moves. It is important to stress that Reidemeister's Theorem only tells us that if two knots are equivalent then there is a sequence of Reidemeister moves exhibiting this, but, it does not tell us what that sequence is. In general, the problem of determining whether two knots are equivalent is quite difficult.

One method for distinguishing between different knots is to assign what are known as invariants to knots. As their name implies, invariants in mathematics are invariant under certain transformations. In the case of knots, invariants should be invariant under the Reidemeister moves; in other words, an invariant of a knot should not change its value when a Reidemeister move is applied to a knot. If this is the case, the Reidemeister's Theorem immediately implies that if two knots are equivalent then they have the same invariant. Notice however, that there is no reaon to believe the converse, knots with the same invariant need not be equivalent. An easy way to see this is to assign to all knots the number 1. This is clearly an invariant, but it is not very useful since it does not help use distinguish any pair of knots. The question of finding useful invariants though, is of considerable interest.

One such invariant is known as the the colouring invariant of a knot. To explain how this invariant is defined suppose that a projection of a knot is given. It has already been observed that if there are k crossings in this projection, then these also decompose the string into k disjoint arcs. The integer m is said to be a colouring invariant of this knot if it is possible to colour the arcs using at least two different integers (in other words, assign to each arc an integer and so that at least two arcs are given different integers) in such a way that at any crossing, if a and b are the integers assigned to the two arcs which end (or begin, depending on your point of view) at the crossing and c is the colour of the arc which passes on top at the crossing then a + b = 2c mod m. The mod m is very important here because it allow knots to be classified into several distinct equivalence classes. Without this, there would be only two possibilities: either the knot can be coloured or not. This would mean that knots are divided into only two classes. Using colouring mod m allows a much finer analysis. Finally, notice that the requirement that at least two colours are used is also crucial since any knot can be coloured trivially using only one colour so that all the equations are satisfied.

Of course simply calling something an invariant does not make it an invariant. It is necessary to supply a proof that the colouring invariant is an invariant before anything further can be done with it. However, before looking at this proof, it is probably instructive to look at an example of a knot and a colouring of it. In the projection shown, the knot at the left has 4 arcs and, of courses, 4 crossings as well. The crossings have been labeled a , b , c and d while the arcs have been coloured with integers from 0 to 2. Each of the crossings corresponds to an equation which must be satisfied by any colouring of the arcs. For example, at crossing a the arc passing below from the bottom is coloured 0, the arc passing below from the top is coloured 2 and the arc passing on top of the crossing is coloured 1. At this crossing, the equation 0 + 2 = 1?2 mod m must be satisfied. But what is the value of m? It turns out that in this case m = 3 will work. Since 2 = 1?2 mod m for any value of m the equation at crossing a is certainly satisfied. However, the equation at crossing b uses arithmetic modulo 3 in an essential way. the equation at this crossing is 0 + 1 = 2?2 mod m and it is satisfied mod m only if m=3. The equation at crossing c is also 0 + 1 = 2?2 mod 3 while the equation at crossing d is 0 + 2 = 2?1 mod 3.

Use the colouring invariant to determine whether or not the following two knots are equivalent.

When looking over the provided solution pay attention to the fact that the solution is a not perfect but this fcat is not hidden and is mentioned in the summary.


Mike Zabrocki
Email address:
Department of Mathematics and Statistics
York University
Back to Department's Public Page
Back to my course Page
Back to top of page