next up previous contents
Next: Quandle -Cocycle Invariants of Up: Quandle Cocycle Invariants of Previous: Quandle Cocycle Invariants of   Contents

Quandle $ 2$-Cocycle Invariants of Classical Knots

The cocycle invariant for classical knots [CJKLS03] using quandle $ 2$-cocycles was defined as follows. Let $ X$ be a finite quandle and $ A$ an abelian group.

A function $ \phi : X \times X \rightarrow A$ is called a quandle $ 2$-cocycle if it satisfies the $ 2$-cocycle condition

$\displaystyle \phi(x,y) - \phi(x,z) + \phi(x*y, z) - \phi(x*z, y*z)=0, $

for all $ x,y,z \in X$ and $ \phi(x,x)=0$ for all $ x \in X$. Let $ {\mathcal C}$ be a coloring of a given diagram $ D$ of a knot $ K$ by $ X$. The $ ($Boltzmann$ )$ weight $ B( {\mathcal C}, \tau)$ at a crossing $ \tau$ of $ K$ is then defined by $ B( {\mathcal C}, \tau)=\phi(x_{\tau}, y_{\tau})^{\epsilon(\tau)}$, where $ (x_{\tau}$, $ y_{\tau})$ is the ordered pair of colors (source colors) at $ \tau$ and $ \epsilon(\tau)$ is the sign ($ \pm 1$ for positive and negative crossings, respectively) of $ \tau$. Here $ B( {\mathcal C}, \tau)$ is an element of $ A$ written multiplicatively. In Fig. [*] of Chapter $ 2$, if the under-arc of the crossing $ \tau$ depicted is oriented downwards from $ \alpha$ to $ \gamma$, then $ \epsilon(\tau)=1$, and $ B( {\mathcal C}, \tau)=\phi(a,b)$ with the indicated coloring. If the orientation of the under-arc is opposite, then $ B( {\mathcal C}, \tau)=\phi(a,b)^{-1}$.

Definition 4.1 ([CJKLS03])   The formal sum (called a state-sum) in the group ring $ \mathbb{Z}[A]$

$\displaystyle \Phi_{\phi} (K) = \sum_{{\mathcal C}\in {\rm Col_X(D)}} \prod_{\tau}
B( {\mathcal C}, \tau ) $

is called the quandle $ 2$-cocycle invariant.

The value (an element of $ \mathbb{Z}[A]$) of the invariant is written as $ \sum_{g \in G} a_g g $, so that if $ A=\mathbb{Z}_n=\{ u^k  \vert  0 \leq k < n \}$, the value is written as a polynomial $ \sum_{k=0}^{n-1} a_k u^k$. The fact that $ \Phi_{\phi} (K) $ is a knot invariant is proved easily by checking Reidemeister moves [CJKLS03]. For example, in Fig. [*] of Chapter $ 2$, the product of weights that appear in the LHS is $ \phi(x,y) \phi(x*y, z) \phi(y,z)$ from the top to bottom crossings, and for the RHS it is $ \phi(y,z)\phi(x,z) \phi(x*z, y*z)$, and the equality obtained is exactly the $ 2$-cocycle condition in multiplicative notation, after canceling $ \phi(y,z)$.

The cocycle invariant can be also written as a family (multi-set, a set with repetition allowed) of weight sums [Lop03]

$\displaystyle \left\{ \left. \sum_{\tau} B_A( {\mathcal C}, \tau )  \right\vert
 {\mathcal C}\in {\rm Col_X(K)} \right\} $

where now the values of $ B$ in $ A$ are denoted by additive notation, $ B_A( {\mathcal C}, \tau ) = {\epsilon(\tau) } \phi(x_{\tau}, y_{\tau}) $. Then this is well-defined for any quandle of any cardinality. The sum of the value of the weight for a specific coloring is also called the contribution of the coloring to the invariant.

The $ 2$-cocycle invariant is well-defined for virtual knots, as checked in a similar manner.

Example 4.2   Consider a trefoil $ K$ represented by the closure of the braid $ (\sigma_1)^3$, where we denote the standard braid generator by $ \sigma_i$, which is represented by a positive crossing placed between $ i\/$th and $ (i+1)\/$st strings.

Let $ X=A=\mathbb{Z}_2[t]/(t^2+t+1)$, an Alexander quandle as $ X$ and an abelian group as $ A$, and put the source colors $ (1,0)$, for example, at the top crossing. This color vector goes down to the next crossing, giving $ (0, t)$, $ (t, t(1+t))$=$ (t, 1)$, and comes back to $ (1, t^2 + t+1)=(1,0)$, by the coloring rule, and defines a non-trivial coloring $ {\mathcal C}$. In fact any pair extends to a coloring, and $ \vert{\rm Col}_X(K)\vert=4^2$.

Consider the function $ \phi : X \times X \rightarrow A$ defined by $ \phi(x,y)=(x-y)^2 y $. Then this satisfies the $ 2$-cocycle condition by calculation (we will further discuss this method of construction of $ 2$-cocycles later).

The contribution is computed by

$\displaystyle \sum_{\tau} B_A( {\mathcal C}, \tau )= 0 + (-t)^2 t + (t-1)^2 = t+1 \in A,$

and after we compute it for all colorings, we obtain, as a multiset,

$\displaystyle \Phi_\phi(K)=\{ \sqcup_4 (0) , \sqcup_{12} (t+1) \} ,$

where $ \sqcup_n x$ denotes $ n$ copies of $ x$ for a positive integer $ n$.

We make the following notational convention. In the above example, if we denote additive generators $ 1$ and $ t$ multiplicatively by $ h$ and $ k$, then the elements of $ X$ in multiplicative notation are written as $ \{ 1, h, k, hk \}$. Hence the invariant in the state-sum form is written as $ \Phi_\phi (K)=4 + 12 hk$. In Maple calculations, it is convenient and easier to see outputs if we use the symbols $ u(=u^1)$ and $ u^t$ for $ h$ and $ k$. Then the additive notations remain on superscripts and the addition is also retained in exponents ( $ hk=u \cdot u^t=u^{(t+1)}$). With this convention the invariant value is written as $ 4+12 u^{(t+1)}$. We use this notation if there is no confusion.


next up previous contents
Next: Quandle -Cocycle Invariants of Up: Quandle Cocycle Invariants of Previous: Quandle Cocycle Invariants of   Contents
Masahico Saito - Quandle Website 2006-09-19