---
title: Kauffman 4-Strand Diagram Monoid
url: https://www.emergentmind.com/topics/kauffman-4-strand-diagram-monoid
type: topic
---

# Kauffman 4-Strand Diagram Monoid

The Kauffman 4-strand diagram monoid is the 4-strand specialization of the planar Kauffman–Temperley–Lieb diagram calculus. In semigroup-theoretic form it is the monoid \(\mathcal K_4=\mathbb N\times \mathcal J_4\), where \(\mathcal J_4\) is the Jones monoid of planar non-crossing pairings on four top and four bottom vertices, and the extra \(\mathbb N\)-coordinate records the number of closed middle loops created under composition. In algebraic presentation form it is generated by a loop element \(c\) and hooks \(h_1,h_2,h_3\) subject to Temperley–Lieb relations. In computational knot-theoretic work, the same 4-strand planar connectivity types appear as a 14-element basis \(g_1,\dots,g_{14}\) for bracket-state propagation under concatenation [1507.04838; 1910.09190; 2508.17904].

## 1. Definitions and competing conventions

Fix \(n\in\mathbb N\). The Jones or Temperley–Lieb monoid \(\mathcal J_n\) consists of planar Brauer diagrams on top vertices \(1,\dots,n\) and bottom vertices \(1',\dots,n'\), with every block of size \(2\). The Kauffman diagram monoid is then defined as
\[
\mathcal K_n=\mathbb N\times \mathcal J_n,
\]
with a twisted product that adds a loop count to the first coordinate. For \(n=4\), \(\mathcal J_4\) has \(C_4=14\) elements, so the circle-free planar core already has 14 distinct connectivity types [1507.04838].

A parallel algebraic presentation uses generators \(c,h_1,\dots,h_{n-1}\) and relations
\[
h_i h_j=h_j h_i \quad (|i-j|\ge 2),\qquad
h_i h_j h_i=h_i \quad (|i-j|=1),\qquad
h_i^2=ch_i=h_ic.
\]
For \(n=4\), this gives generators \(c,h_1,h_2,h_3\). Diagrammatically, this presented monoid is isomorphic to the planar connection monoid on four strands with circles, so the loop generator \(c\) and the \(\mathbb N\)-coordinate in \(\mathbb N\times\mathcal J_4\) encode the same loop-counting phenomenon in different languages [1602.01157; 1405.0783].

In the 4-tangle bracket literature, the phrase “Kauffman 4-strand diagram monoid” is used operationally for a fixed basis of 14 planar 4-strand states \(g_1,\dots,g_{14}\), with loops absorbed into coefficients in \(\mathbb Z[x]\) rather than kept as separate monoid data. In generalized diagram-category language, the same 4-strand object appears as the Temperley–Lieb/Kauffman monoid \(TL_\delta(4)\), again with 14 planar basis diagrams [2508.17904; 2512.17177].

| Realization | Elements in the 4-strand case | Treatment of loops |
|---|---|---|
| Twisted semigroup \(\mathcal K_4\) | Pairs \((i,\alpha)\) with \(\alpha\in\mathcal J_4\) | Stored in the \(\mathbb N\)-coordinate |
| Presented monoid \(K_4\) | Words in \(c,h_1,h_2,h_3\) modulo TL relations | Stored by powers of \(c\) |
| State basis \(\{g_1,\dots,g_{14}\}\) | 14 circle-free planar states | Absorbed into coefficients in \(\mathbb Z[x]\) |
| Categorical \(TL_\delta(4)\) | 14 planar matchings | Each floating loop evaluates to \(\delta\) |

This suggests that most differences in usage concern bookkeeping rather than the underlying planar 4-strand connectivity data.

## 2. Diagrammatics and multiplication

A 4-strand Jones diagram has top vertices \(1,2,3,4\) and bottom vertices \(1',2',3',4'\). Its pairs may be upper hooks \(\{i,j\}\), lower hooks \(\{i',j'\}\), or transversals \(\{i,j'\}\), subject to planarity. Typical examples are the identity diagram
\[
\{1,1'\},\{2,2'\},\{3,3'\},\{4,4'\}
\]
and the rank-\(0\) hook diagram
\[
\{1,2\},\{3,4\},\{1',2'\},\{3',4'\}.
\]
For \(n=4\), possible ranks are \(0,2,4\) [1507.04838].

Multiplication is defined by stacking diagrams. If \(\alpha,\beta\in\mathcal J_4\), one composes them by placing \(\alpha\) above \(\beta\), identifying the middle row, and tracing the resulting planar connections. In the Kauffman monoid this is twisted by the number \(\tau(\alpha,\beta)\) of floating middle loops:
\[
(i,\alpha)\star(j,\beta)=\big(i+j+\tau(\alpha,\beta),\alpha\beta\big).
\]
The second coordinate is the ordinary Jones product; the first coordinate accumulates loop count. This is the standard loop-counting twist corresponding to the exponent of the Temperley–Lieb loop parameter in the algebra product \(\alpha\circ\beta=\xi^{\tau(\alpha,\beta)}\alpha\beta\) [1507.04838].

The same composition law appears in the connection-monoid model. There one works with left pins \(1,2,3,4\), right pins \(1',2',3',4'\), t-wires, l-wires, r-wires, and circles. Multiplication glues right pins of the first diagram to left pins of the second, then traces outer connections and counts new circles. The identity is the diagram with straight through wires \(i-i'\), and each hook \(h_i\) joins \(i\) to \(i+1\) on one side and \(i'\) to \((i+1)'\) on the other [1910.09190].

In the 4-tangle state-space approach, one fixes 14 basis diagrams \(g_1,\dots,g_{14}\). Any smoothed 4-tangle state is expressed as a linear combination of these basis states, and concatenation of tangles induces multiplication through the concatenation table of the \(g_i\). The identity basis state is the straight-through 4-strand diagram, represented by the initial vector
\[
A_0=[1,0,\dots,0]^\intercal
\]
in transfer-matrix calculations [2508.17904].

## 3. Idempotents, interface graphs, and the 4-strand count

An element \(e\) of a monoid is idempotent if \(e^2=e\). In the twisted Kauffman monoid this becomes
\[
(i,\alpha)\star(i,\alpha)=(i,\alpha),
\]
hence
\[
(2i+\tau(\alpha,\alpha),\alpha^2)=(i,\alpha).
\]
Therefore an idempotent must satisfy
\[
i=0,\qquad \tau(\alpha,\alpha)=0,\qquad \alpha^2=\alpha.
\]
The classification in planar diagram monoids is phrased in terms of the interface graph \(\Gamma_\alpha\) associated to a diagram \(\alpha\) [1507.04838].

For \(\alpha\in\mathcal M_n\), and in particular for \(\alpha\in\mathcal J_n\), the interface graph \(\Gamma_\alpha\) has vertex set \(1,\dots,n\). Upper hooks give edges colored \(-1\), lower hooks give edges colored \(+1\), and each vertex \(v\) receives a color \(c(v)\in\mathbb Z_2\times\mathbb Z_2\) recording whether \(v\) belongs to the codomain and domain of transversals. Because the original diagram is planar, every connected component of \(\Gamma_\alpha\) is either a cycle or a path, with alternating edge colors. Paths are classified as active, inactive, or mixed according to endpoint colors.

For Jones idempotents, every component of \(\Gamma_\alpha\) must be either a cycle or an active path of even length. For Kauffman idempotents, the twist imposes the stronger condition that every component must be an active path of even length; cycles and inactive paths are excluded because they create floating middle loops. Corollary 3.7 states:
\[
(i,\alpha)\in E(\mathcal K_n)\iff i=0\ \text{and every component of }\Gamma_\alpha\text{ is an active path of even length}.
\]
Equivalently, Kauffman idempotents are precisely those Jones idempotents whose interface graphs have no cycle components [1507.04838].

The paper gives an explicit enumeration formula
\[
|E(\mathcal K_n)|
=\sum_{\alpha\in\Delta(\mathcal J_n)\atop \Xi(\alpha)=\Theta(\alpha)}
\prod_{\theta\in\Theta(\alpha)}u_\theta(\alpha)\,l_\theta(\alpha),
\]
where \(\Theta(\alpha)\) records cycle components containing both upper and lower outer hooks, and \(u_\theta(\alpha),l_\theta(\alpha)\) count the corresponding outer hooks. For \(n=4\), the outcome is
\[
|E(\mathcal K_4)|=5.
\]
One of these five is the identity diagram. Its interface graph has four isolated vertices, each colored \(\begin{bmatrix}1\\1\end{bmatrix}\), so every component is an active path of even length \(0\). The remaining four idempotents are not listed explicitly in the paper, but the general theory shows that each is of the form \((0,\alpha)\) with a cycle-free active-path interface graph [1507.04838].

## 4. Presentations, normal forms, idempotent generation, and identities

The presented Kauffman monoid \(K_n\) admits a canonical Jones normal form. Writing
\[
h[j,i]=h_j h_{j-1}\cdots h_i,
\]
every element has a unique form
\[
w=c^\ell h[b_1,a_1]\cdots h[b_k,a_k]
\]
with \(k,\ell\ge 0\) and strictly increasing sequences \(a_1<\cdots<a_k\) and \(b_1<\cdots<b_k\). For \(n=4\), the available indices are \(1,2,3\), so all normal forms are built from the six blocks \(h[1,1],h[2,1],h[2,2],h[3,1],h[3,2],h[3,3]\) and powers of \(c\) [1602.01157].

A key refinement colors each block by parity. A block is white if its endpoints have different parity, blue if both endpoints are odd, and red if both are even. If \(\blue{w}\), \(\red{w}\), and \(\cc{w}\) denote the numbers of blue blocks, red blocks, and occurrences of \(c\) in the normal form of \(w\), then the characteristic number is
\[
\chi(w)=\cc{w}-|\blue{w}-\red{w}|.
\]
The main criterion is
\[
w\in\langle E_n\rangle \iff \chi(w)\ \text{is non-negative and even},
\]
where \(\langle E_n\rangle\) is the idempotent-generated subsemigroup. For \(n=4\),
\[
\mathrm{rank}(\langle E_4\rangle)=4,\qquad \mathrm{idrank}(\langle E_4\rangle)=4.
\]
A minimal idempotent generating set is provided by the four length-\(2\) blocks and inverse blocks
\[
h[2,1],\ h[3,2],\ h[1,2],\ h[2,3]
\]
in this small case [1602.01157].

The equational theory of the 4-strand monoid is unusually rigid. The identities of \(\mathcal K_n\) are nonfinitely based for every \(n\ge 3\), and this remains true when \(\mathcal K_n\) is regarded as an involution semigroup under either natural involution. Thus \(\mathcal K_4\) has no finite identity basis [1405.0783].

At the same time, \(\mathcal K_4\) admits an unexpectedly tractable identity theory. The monoids \(\mathcal K_3\) and \(\mathcal K_4\) satisfy exactly the same identities. An identity \(w=w'\) holds in \(\mathcal K_4\) if and only if \(\operatorname{alph}(w)=\operatorname{alph}(w')\) and, for every subset \(Y\subseteq \operatorname{alph}(w)\), the words \(u=w_Y\) and \(u'=w'_Y\) obtained by deleting all letters in \(Y\) have the same first letter, the same last letter, and the same number of occurrences of every word of length \(2\). This yields a polynomial-time identity-checking algorithm for \(\mathcal K_4\), with time \(O(kn\log(kn))\) when \(|\operatorname{alph}(w)|=k\) and \(|ww'|=n\) [1910.09190].

A common misconception is that nonfinite basability should imply intractable identity checking. Here the two phenomena coexist: finite axiomatizability fails, but the decision problem remains polynomial-time.

## 5. Finite-state transfer algebra for 4-tangles

In the recursive computation of bracket polynomials for iterated 4-tangles, the Kauffman 4-strand diagram monoid functions as a finite state space. A 4-tangle shadow \(T\) is expanded as
\[
T=\sum_{i=1}^{14} a_i g_i,
\]
where \(g_1,\dots,g_{14}\) is a fixed basis of 14 elementary planar 4-strand diagrams and \(a_i\in\mathbb Z[x]\). Concatenation of tangles induces a linear transformation on the coefficient vector
\[
A(T)=\begin{bmatrix}a_1\\ a_2\\ \vdots\\ a_{14}\end{bmatrix}.
\]
If \(T_n=T^n\), then
\[
A_n=MA_{n-1}=M^nA_0,
\]
where \(M\) is a \(14\times 14\) states matrix with entries in \(\mathbb Z[x]\) and
\[
A_0=[1,0,\dots,0]^\intercal
\]
corresponds to the identity state [2508.17904].

Closure turns the state vector into a bracket polynomial. In one formulation, a closure-weight vector
\[
\overline X=\left[x^4,x^3,x^3,x^3,x^2,x^2,x^2,x^2,x,x^2,x^2,x,x,x\right]^\intercal
\]
gives
\[
\langle \overline{T_n}\rangle = A_n\cdot \overline X.
\]
Thus the entire bracket recursion is reduced to repeated multiplication by a fixed \(14\times 14\) matrix. For the square Turk’s head generator, the bracket expansion has coefficients
\[
a_1=a_2=a_3=a_4=a_5=a_8=a_9=a_{11}=1,
\]
with all other \(a_i=0\) [2508.17904].

The same 14-state framework appears in the Celtic-link computation. There the fundamental tangle \(G\) has
\[
\langle G\rangle
=
\langle g_1\rangle+\langle g_2\rangle+\langle g_3\rangle+\langle g_4\rangle+\langle g_5\rangle+\langle g_8\rangle+\langle g_9\rangle+\langle g_{11}\rangle,
\]
and the associated \(14\times 14\) state matrix \(M\) has characteristic polynomial
\[
\chi(M,\lambda)
=
(\lambda-1)(\lambda-x-1)^3(\lambda-x-2-r)^3(\lambda-x-2+r)^3
\left(\lambda-\frac{p-q}{2}\right)^2
\left(\lambda-\frac{p+q}{2}\right)^2
\]
with
\[
r=\sqrt{2x+3},\qquad p=x^2+4x+4,\qquad q=\sqrt{x^4+4x^3+12x^2+20x+12}.
\]
After closure,
\[
\langle \overline{G_n}\rangle=(x+1)^2\langle CK_4^{2n}\rangle,
\]
and the matrix-eigenvalue analysis yields a closed form for \(\langle CK_4^{2n}\rangle\) [2508.10410].

The two bracket papers use the same 14-state philosophy but different closure data. This indicates that basis orderings and closure conventions vary across the literature, even when the underlying 4-strand state space is the same.

## 6. Skein algebras, character varieties, and generalized diagram categories

The phrase “Kauffman 4-strand diagram monoid” does not appear explicitly in the study of the Kauffman bracket skein algebra of the 4-holed disk, but the natural matching object is the diagrammatic multiplicative structure of
\[
\mathcal S_4=\mathcal S(\Sigma_{0,5};\mathbb Z[q^{\pm \frac12}]),
\]
where multiplication is stacking. The generators are simple closed curves
\[
t_{i_1\cdots i_r},
\]
indexed by nonempty subsets of \(\{1,2,3,4\}\). The algebra admits a monomial basis over \(\mathbb Z[q^{\pm \frac12}][t_1,t_2,t_3,t_4]\) consisting of words
\[
t_{12}^{j_1}t_{23}^{j_2}t_{34}^{j_3}t_{14}^{j_4}a
\]
with \(a\) in an explicitly listed set of heavy generators involving \(t_0,t_{13},t_{24},t_{123},t_{124},t_{134},t_{234}\). In the classical limit \(q^{1/2}=-1\), these diagram generators map to trace functions on the \(\mathrm{SL}(2,\mathbb C)\)-character variety of the rank-4 free group \(F_4\) [2411.15829].

A different skein-theoretic extension appears in the exterior of a 4-strand Montesinos knot. There the boundary diagram algebra acts on the skein module of a handlebody, and passing to the knot exterior imposes relations identifying the four boundary meridians. The resulting skein module contains nonzero torsion: there exists \(e\neq 0\) with
\[
(q^4-1)e=0,
\]
and more generally a family \(e^{(n)}\neq 0\) with
\[
(q^{2n}-1)e^{(n)}=0.
\]
This provides a negative answer to Kirby’s Problem 1.92 (G)-(i) in that 4-strand Montesinos setting [2311.01177].

In the 2025 generalized-cobordism framework, the classical Kauffman object is identified with the Temperley–Lieb/Kauffman monoid \(TL_\delta(4)\). It has 14 elements partitioned into three \(\mathcal D\)-classes \(J_0,J_2,J_4\) according to through-strand counts \(0,2,4\), so the decomposition is \(4+9+1\). In the generic semisimple regime, the corresponding simple modules have dimensions \(2,3,1\). The same paper shows that tight twistings preserve Green-theoretic cell structure in a controlled way, so Kauffman-type loop-counting twists are representation-theoretically mild [2512.17177].

Taken together, these developments place the Kauffman 4-strand diagram monoid at an intersection of semigroup theory, transfer-matrix methods, skein algebras, and categorical representation theory. Its smallest nontrivial 4-strand instance already exhibits infinite loop-counting structure, a finite 14-state planar core, nontrivial idempotent combinatorics, nonfinite equational basis behavior, polynomial-time identity checking, and direct applications to bracket-polynomial recursion.

Source: https://www.emergentmind.com/topics/kauffman-4-strand-diagram-monoid