\documentclass[11pt,a4paper]{article}

\usepackage[utf8]{inputenc}
\usepackage[T1]{fontenc}
\usepackage{lmodern}
\usepackage{amsmath,amssymb,amsthm}
\usepackage{graphicx}
\usepackage[margin=28mm]{geometry}
\usepackage{xurl}
\usepackage[hidelinks]{hyperref}
\usepackage{microtype}

\newtheorem{theorem}{Theorem}[section]
\newtheorem{proposition}[theorem]{Proposition}
\newtheorem{lemma}[theorem]{Lemma}
\newtheorem{corollary}[theorem]{Corollary}
\newtheorem{conjecture}[theorem]{Conjecture}
\theoremstyle{definition}
\newtheorem{definition}[theorem]{Definition}
\newtheorem{remark}[theorem]{Remark}
\newtheorem{example}[theorem]{Example}

\newcommand{\Q}{\mathbb{Q}}
\newcommand{\Z}{\mathbb{Z}}
\newcommand{\N}{\mathbb{N}}
\newcommand{\g}{\mathfrak{g}}
\newcommand{\slt}{\mathfrak{sl}(2)}
\newcommand{\Ue}{U(\g)}
\newcommand{\tr}{\operatorname{tr}}
\newcommand{\cro}{\operatorname{cr}}
\newcommand{\cyc}{\operatorname{cyc}}
\newcommand{\des}{\operatorname{des}}
\newcommand{\dd}{\partial}
\newcommand{\binomset}[2]{\tbinom{#1}{#2}}

\title{The $\mathfrak{sl}(2)$ weight system at $c=-3/32$ extends to a graph 4-invariant}

\author{Deep Bhattacharjee\thanks{Corresponding author. Email: \href{mailto:d.bhattacharjee@erl-forschung.de}{d.bhattacharjee@erl-forschung.de}, \href{mailto:itsdeep@live.com}{itsdeep@live.com}, \href{mailto:egspl@proton.me}{egspl@proton.me}. ORCID: \href{https://orcid.org/0000-0003-0466-750X}{0000-0003-0466-750X}.}\\[2pt]
\normalsize Formerly, Electro-Gravitational Space Propulsion Laboratory (EGSPL),\\
\normalsize Bhubaneswar, Odisha 751030, India\\[1pt]
\normalsize Formerly, University of Calcutta, Senate House, 87/1 College Street,\\
\normalsize Kolkata 700073, West Bengal, India
\and
Ushashi Bhattacharya\thanks{Email: \href{mailto:rai.bhattacharya.in@gmail.com}{rai.bhattacharya.in@gmail.com}. ORCID: \href{https://orcid.org/0009-0002-2254-3914}{0009-0002-2254-3914}.}\\[2pt]
\normalsize Formerly, National Taiwan University,\\
\normalsize Taipei 106319, Taiwan}

\date{26 October 2026}

\begin{document}

\maketitle

\begin{abstract}
\itshape
A weight system is a function on chord diagrams that satisfies the four-term relations. The Lie algebra $\mathfrak{sl}(2)$ gives a weight system with values in the polynomial ring $\mathbb{Q}[c]$, and its value on a chord diagram depends only on the intersection graph of the diagram. Fomichev, Karev, Pavutnitskiy and Usanov showed that this weight system does not come from a function on graphs, and that a specialisation at a value $c=c_0$ can come from one only for $c_0=0$, $3/8$, $1$ or $-3/32$. The first three values were settled before. They conjectured that the last one, $c=-3/32$, which is the eigenvalue of the Casimir element on the oscillator representation, also extends. We prove this. The extension is the function $\xi(G)=(-32)^{-n}S_G(-2)$, where $G$ is a graph with $n$ vertices and $S_G(N)$ is a signed count of the maps that send each vertex of $G$ to a two-element subset of $\{1,\dots,N\}$: a map counts with the sign $-1$ for every edge whose two endpoints are sent to subsets meeting in exactly one element. The count is a polynomial in $N$, so its value at $N=-2$ makes sense. The four-term relation for $\xi$ is proved by an involution on these maps. The identity with the weight system is proved by letting the chord diagram act on Verma modules, then on the oscillator representation, where a supertrace on the Weyl algebra turns the value into a sum over perfect matchings, and finally by reading the same sum as a trace in a Clifford algebra on $N$ generators.
\end{abstract}

\medskip
\noindent\textbf{Keywords.} weight system; intersection graph; four-term relation; Clifford algebra.

\smallskip
\noindent\textbf{MSC 2020.} 05C31; 57K16; 17B10; 15A66.

\medskip
\begin{center}
\begin{minipage}{0.92\textwidth}
\small\sloppy
\noindent\textbf{Note.} The earlier versions and their checking programs are archived on Zenodo under the concept DOI \href{https://doi.org/10.5281/zenodo.23312529}{10.5281/zenodo.23312529} (versions \href{https://doi.org/10.5281/zenodo.23312664}{10.5281/zenodo.23312664}, \href{https://doi.org/10.5281/zenodo.23384025}{10.5281/zenodo.23384025}, \href{https://doi.org/10.5281/zenodo.23541142}{10.5281/zenodo.23541142} and \href{https://doi.org/10.5281/zenodo.23630528}{10.5281/zenodo.23630528}). The source of this revision is in the repository \url{https://github.com/creelie/Paper_7}, release \texttt{v1.0.1}, archived on Zenodo as \href{https://doi.org/10.5281/zenodo.23658136}{10.5281/zenodo.23658136} (concept DOI \href{https://doi.org/10.5281/zenodo.23658071}{10.5281/zenodo.23658071}).
\end{minipage}
\end{center}

\section{Introduction}\label{sec:intro}

A weight system is a function on chord diagrams that satisfies the four-term relations. Weight systems are the symbols of finite type knot invariants \cite{Kon93,BN95,CDM12}.\footnote{A chord diagram with $n$ chords is a circle with $2n$ marked points joined in pairs. A finite type invariant of degree $n$ of knots, evaluated on a knot with $n$ double points and extended by the Vassiliev skein relation, depends only on the chord diagram of the double points; the function obtained in this way satisfies the four-term relations, and by Kontsevich's theorem every function satisfying them arises so.} A Lie algebra with an invariant form gives a weight system with values in the centre of its universal enveloping algebra. For $\slt$ with the Killing form the centre is the polynomial ring $\Q[c]$ in the Casimir element $c$, and Chmutov and Lando proved that the value $w_D\in\Q[c]$ of this weight system on a chord diagram $D$ depends only on the intersection graph $\Gamma(D)$ of $D$ \cite{ChL07}. Lando's four-term relations for graphs \cite{Lan00} are such that every function $f$ on graphs satisfying them gives a weight system $D\mapsto f(\Gamma(D))$; such a function is called a $4$-invariant. Lando asked whether $w$ comes from a $4$-invariant \cite{ChL07}. For graphs with at most $8$ vertices this is true \cite{Kra21}; see \cite{KL22} for a survey.

Fomichev, Karev, Pavutnitskiy and Usanov \cite{FKPU} answered the question in the negative. They found a linear combination of intersection graphs that vanishes modulo the four-term relations of graphs and on which $w$ takes the value $c(c-\frac38)(c-1)(c+\frac{3}{32})$. So the specialisation of $w$ at a value $c=c_0$ can extend to a $4$-invariant only if $c_0$ is one of $0$, $\frac38$, $1$ and $-\frac{3}{32}$. The first three values are the eigenvalues of the Casimir element on the irreducible representations of dimension $1$, $2$ and $3$, and for them extensions are known \cite{FKPU,FK25,YDJ26}.\footnote{At $c=0$ the weight system is identically $1$ on diagrams, since the Casimir element acts by $0$ on the trivial representation. At $c=\frac38$ the extension was conjectured in \cite{FK25} and proved in \cite{YDJ26}. At $c=1$ it is given in \cite{FKPU}.} The value $-\frac{3}{32}$ is the eigenvalue of the Casimir element on the oscillator representation, the representation of $\slt$ on polynomials in one variable by second order differential operators. The authors of \cite{FKPU} built an extension for graphs with at most $10$ vertices and stated the following conjecture.

\begin{conjecture}[{\cite[Section~4]{FKPU}}]\label{conj:fkpu}
The specialisation of the $\slt$-weight system at $c=-\frac{3}{32}$ extends to a $4$-invariant of graphs $\xi(G)$.
\end{conjecture}

We prove Conjecture~\ref{conj:fkpu}, with an explicit formula for $\xi$. All graphs are finite and simple. For $N\in\N$ write $[N]=\{1,\dots,N\}$ and $\binomset{[N]}{2}$ for the set of its $2$-element subsets.

\begin{definition}\label{def:S}
For a graph $G=(V,E)$ and $N\in\N$ put
\[
S_G(N)=\sum_{p\colon V\to\binomset{[N]}{2}}\ \prod_{xy\in E}(-1)^{|p(x)\cap p(y)|},
\]
and for $k\in\N$ let $T_k(G)$ be the same sum restricted to the maps $p\colon V\to\binomset{[k]}{2}$ with $\bigcup_{x\in V}p(x)=[k]$.
\end{definition}

\begin{theorem}\label{thm:main}
Let $G=(V,E)$ be a graph with $n$ vertices.
\begin{enumerate}
\item[(a)] $S_G(N)=\sum_{k=0}^{2n}\binom{N}{k}T_k(G)$ for all $N\in\N$; so $S_G$ is a polynomial in $N$ of degree at most $2n$.
\item[(b)] The graph invariant
\[
\xi(G)=\frac{S_G(-2)}{(-32)^{n}}=\frac{1}{(-32)^{n}}\sum_{k=0}^{2n}(-1)^k(k+1)\,T_k(G)
\]
satisfies the four-term relations of graphs.
\item[(c)] For every chord diagram $D$, $w_D(-\frac{3}{32})=\xi(\Gamma(D))$.
\end{enumerate}
\end{theorem}

Parts (b) and (c) are Conjecture~\ref{conj:fkpu}. The two expressions for $\xi$ agree by (a), because $\binom{-2}{k}=(-1)^k(k+1)$.

\begin{example}\label{ex:small}
For one vertex, $S_{K_1}(N)=\binom N2$ and $\xi(K_1)=3/(-32)=-\frac{3}{32}$, the value of $w$ on one chord. Since $S_{G\sqcup H}=S_GS_H$, $\xi$ is multiplicative, as $w$ is. For two adjacent vertices, the pairs $(p(x),p(y))$ with $|p(x)\cap p(y)|=2,1,0$ number $\binom N2$, $\binom N2\cdot 2(N-2)$ and $\binom N2\binom{N-2}{2}$, so $S_{K_2}(N)=\binom N2\bigl(1-2(N-2)+\binom{N-2}{2}\bigr)$ and $\xi(K_2)=3\cdot19/1024=\frac{57}{1024}$. This is $c(c-\frac12)$ at $c=-\frac{3}{32}$, the value of $w$ on two crossing chords.\footnote{On two crossing chords $C_D=C^2-\frac12 C$ in the notation of Section~\ref{sec:verma}; this is the standard computation for the $\slt$ weight system, see \cite[Section 6.1]{CDM12}, with the Killing form normalisation that gives the value $c$ on one chord.}
\end{example}

\begin{figure}[ht]
\centering
\includegraphics[width=0.88\textwidth]{figures/fig_chords.png}
\caption{A chord diagram with three chords, cut between the points $6$ and $1$, and its intersection graph, a path. The map $p$ sends each vertex to a $2$-element subset of $[3]$; both edges meet in one element, so $p$ contributes $(-1)(-1)=1$ to $S_{\Gamma(D)}(3)$.}\label{fig:chords}
\end{figure}

The proof of (b) is an involution on assignments (Section~\ref{sec:proof}). For (c) we let the element $C_D\in U(\slt)$ of a chord diagram act on the Verma module of highest weight $\mu$; it acts by a polynomial $\lambda_D(\mu)$, and $w_D(-\frac{3}{32})=\lambda_D(-\frac12)$ (Section~\ref{sec:verma}). The oscillator representation on $\Q[z]$ is the sum of the Verma modules of highest weights $-\frac12$ and $-\frac32$, so the image of $C_D$ in the Weyl algebra is the scalar $\lambda_D(-\frac12)$. A supertrace on the Weyl algebra with a Wick formula writes this scalar as a sum over perfect matchings, in which every cycle contributes a factor $-2$ (Sections~\ref{sec:osc} and~\ref{sec:matching}). With $N$ in place of $-2$ the same sum is a trace in the Clifford algebra on $N$ generators, and that trace equals $4^nS_\Gamma(N)$ (Section~\ref{sec:clifford}). Figure~\ref{fig:flow} shows the two halves of the argument.\footnote{The step from $N$ to $-2$ is the point of the proof. The Clifford algebra on $N$ generators is a quotient of the Weyl algebra's fermionic counterpart; the Weyl algebra itself behaves like a Clifford algebra on $-2$ generators, because in the Wick formula of Lemma~\ref{lem:tau}(c) every closed cycle of contractions carries the factor $-2$ instead of $N$. This is the familiar sign that supertraces of bosonic systems carry negative dimensions. Nothing in the proof uses this heuristic; the identity is proved as an identity of polynomials in $N$ and then evaluated at $-2$.}

The combinatorial objects of the proof, chord diagrams with their legs, perfect matchings, their cycle and crossing counts, are finite and the proof uses nothing beyond linear algebra over $\Q$ and counting. Related constructions in which a finite complex or a group action is studied through counts of this kind appear in the authors' earlier work \cite{Bha22a,Bha22b,BB26}.

\begin{figure}[ht]
\centering
\includegraphics[width=0.86\textwidth]{figures/fig_flow.png}
\caption{The proof of Theorem~\ref{thm:main}. The right column gives part (b), the left column and the bottom row part (c).}\label{fig:flow}
\end{figure}

\section{The weight system and Verma modules}\label{sec:verma}

\subsection{Chord diagrams and the weight system}
A chord diagram with $n$ chords is a fixed-point-free involution $\sigma$ of $[2n]$; its chords are the pairs $\{p,\sigma(p)\}$. It is a chord diagram on an oriented circle cut at a point that is not an endpoint; cutting at another point rotates $[2n]$ cyclically. Chords $\{p<q\}$ and $\{p'<q'\}$ cross if $p<p'<q<q'$ or $p'<p<q'<q$. The intersection graph $\Gamma(D)$ has the chords as vertices and the crossing pairs as edges; it does not depend on the cut.\footnote{Two chords cross on the circle exactly when their endpoints alternate around the circle, and alternation is unchanged by a cyclic rotation of the labels.}

Let $\g=\slt$ over $\Q$ with basis $E,F,H$, $[H,E]=2E$, $[H,F]=-2F$, $[E,F]=H$, and let
\[
t=\tfrac14\,E\otimes F+\tfrac14\,F\otimes E+\tfrac18\,H\otimes H\in\g\otimes\g
\]
be the Casimir tensor of the Killing form.\footnote{The Killing form of $\slt$ has $\kappa(E,F)=4$ and $\kappa(H,H)=8$, and $t$ is the element dual to it: $t=\sum_i x_i\otimes x^i$ for a basis $x_i$ and its dual basis $x^i$ with respect to $\kappa$. Any other invariant form is a multiple of $\kappa$, and rescaling the form rescales $w_D$ by a power of the scale; the four special values of \cite{FKPU} are stated for the Killing form.} It is symmetric and $\g$-invariant: $(\operatorname{ad}_x\otimes1+1\otimes\operatorname{ad}_x)t=0$ for $x\in\g$. For a chord diagram $D$ let $t_D\in\g^{\otimes 2n}$ be the tensor with a copy of $t$ in the factors $p,\sigma(p)$ of every chord, and let $C_D\in\Ue$ be the image of $t_D$ under $u_1\otimes\dots\otimes u_{2n}\mapsto u_1\cdots u_{2n}$. For one chord $C_D=C=\frac14(EF+FE+\frac12H^2)$.

For $d\ge1$ let $V_d$ be the irreducible representation of dimension $d$, with basis $e_0,\dots,e_{d-1}$ and
\[
Ee_j=j(d-j)\,e_{j-1},\qquad Fe_j=e_{j+1},\qquad He_j=(d-1-2j)\,e_j,
\]
where $e_{-1}=e_d=0$.
The Casimir element $C$ acts on $V_d$ by $c_d=(d^2-1)/8$.\footnote{On the highest weight vector $e_0$ one has $Fe_0=e_1$, $EFe_0=(d-1)e_0$, $FEe_0=0$ and $H^2e_0=(d-1)^2e_0$, so $Ce_0=\frac14\bigl((d-1)+\frac12(d-1)^2\bigr)e_0=\frac{(d-1)(d+1)}{8}e_0$; since $C$ is central and $V_d$ is irreducible, $C$ acts by this scalar everywhere.} Put $W_d(D)=\tr_{V_d}(C_D)/d$. Rotating the cut permutes the factors of $t_D$ cyclically, so $W_d(D)$ depends only on the diagram on the circle.

By Proposition~\ref{prop:wD} below there is a unique $w_D\in\Q[c]$ with $w_D(c_d)=W_d(D)$ for all $d\ge1$. It is the $\slt$-weight system of \cite{ChV97,FKPU}: by Lemma~\ref{lem:central} $C_D$ lies in the centre of $\Ue$, which is $\Q[C]$ \cite{Hum08}, so $C_D=w(C)$ for a polynomial $w$, and taking traces on $V_d$ gives $w(c_d)=W_d(D)$; the weight system of \cite{FKPU} is defined with the Killing form and takes the value $c$ on one chord.

\begin{lemma}\label{lem:central}
$C_D$ commutes with every $x\in\g$.
\end{lemma}

\begin{proof}
$\operatorname{ad}_x$ is a derivation of $\Ue$, so $[x,C_D]$ is the image of $\sum_{p=1}^{2n}(\operatorname{ad}_x)_p\,t_D$, where $(\operatorname{ad}_x)_p$ acts on the factor $p$. Grouping the two endpoints of every chord writes this sum as a sum over chords of tensors containing $(\operatorname{ad}_x\otimes1+1\otimes\operatorname{ad}_x)t=0$.
\end{proof}

\subsection{Verma modules}
Let $R=\Q[\mu]$ and let $M$ be the free $R$-module with basis $v_0,v_1,\dots$ and
\begin{equation}\label{eq:verma}
Ev_j=j(\mu-j+1)\,v_{j-1},\qquad Fv_j=v_{j+1},\qquad Hv_j=(\mu-2j)\,v_j .
\end{equation}
These operators satisfy the relations of $\g$:
\[
(EF-FE)v_j=\bigl((j+1)(\mu-j)-j(\mu-j+1)\bigr)v_j=(\mu-2j)v_j=Hv_j,
\]
and $(HE-EH)v_j=2Ev_j$, $(HF-FH)v_j=-2Fv_j$, because $H$ has eigenvalue $\mu-2j+2$ on $v_{j-1}$ and $\mu-2j-2$ on $v_{j+1}$. So $M$ is a $\Ue$-module on which $\Ue$ acts $R$-linearly. For $\mu_0\in\Q$, specialising $\mu=\mu_0$ gives the Verma module $V(\mu_0)$ of highest weight $\mu_0$.

\begin{proposition}\label{prop:lambda}
There is $\lambda_D\in\Q[\mu]$ with $C_Dm=\lambda_Dm$ for all $m\in M$. Moreover
\begin{enumerate}
\item[(a)] $W_d(D)=\lambda_D(d-1)$ for every $d\ge1$;
\item[(b)] $\lambda_D(\mu)=\lambda_D(-2-\mu)$.
\end{enumerate}
\end{proposition}

\begin{proof}
$H-\mu$ multiplies $v_j$ by $-2j$, so its kernel is $Rv_0$. As $C_D$ commutes with $H$, $C_Dv_0=\lambda_Dv_0$ for some $\lambda_D\in R$, and $C_Dv_j=C_DF^jv_0=F^jC_Dv_0=\lambda_Dv_j$.

(a) In $V(d-1)$, $Ev_d=d(d-1-d+1)v_{d-1}=0$, so $U_d=\operatorname{span}\{v_j:j\ge d\}$ is a submodule. The quotient $V(d-1)/U_d$, with basis the images of $v_0,\dots,v_{d-1}$, is $V_d$ with the formulas above. $C_D$ acts on $V(d-1)$, hence on $V_d$, by $\lambda_D(d-1)$, so $\tr_{V_d}C_D=d\,\lambda_D(d-1)$.

(b) The map $V(-1-d)\to U_d$, $v_j\mapsto v_{d+j}$, is an isomorphism of $\g$-modules: it commutes with $F$; both $Hv_j$ in $V(-1-d)$ and $Hv_{d+j}$ in $V(d-1)$ have eigenvalue $-1-d-2j$; and $Ev_j=j(-d-j)v_{j-1}$ in $V(-1-d)$ while $Ev_{d+j}=(d+j)(-j)v_{d+j-1}$ in $V(d-1)$. Hence $\lambda_D(-1-d)=\lambda_D(d-1)$ for all $d\ge1$, and the polynomial $\lambda_D(\mu)-\lambda_D(-2-\mu)$ has infinitely many roots.\footnote{This is the symmetry of the Harish-Chandra isomorphism for $\slt$: the central character of $V(\mu)$ depends on $\mu$ only through $\mu(\mu+2)$, see \cite[Section 1.10]{Hum08}. The direct argument above avoids the general theory.}
\end{proof}

\begin{proposition}\label{prop:wD}
There is a unique $w_D\in\Q[c]$ with $w_D(c_d)=W_d(D)$ for all $d\ge1$, and $\deg w_D\le n$. It satisfies $\lambda_D(\mu)=w_D\bigl(\mu(\mu+2)/8\bigr)$; in particular
\[
w_D(-\tfrac{3}{32})=\lambda_D(-\tfrac12)=\lambda_D(-\tfrac32).
\]
\end{proposition}

\begin{proof}
By Proposition~\ref{prop:lambda}(b), $\nu\mapsto\lambda_D(\nu-1)$ is even, so $\lambda_D(\nu-1)=g(\nu^2)$ for some $g\in\Q[x]$. Put $w_D(c)=g(8c+1)$. Then $w_D(\mu(\mu+2)/8)=g((\mu+1)^2)=\lambda_D(\mu)$, and $w_D(c_d)=\lambda_D(d-1)=W_d(D)$ by Proposition~\ref{prop:lambda}(a). Uniqueness holds because the $c_d$ are pairwise distinct. Each term of $C_Dv_0$ is obtained by applying $2n$ operators from \eqref{eq:verma}; their coefficients have degree at most one in $\mu$. So $\deg\lambda_D\le2n$. Translation by $1$ preserves this degree bound, so $\lambda_D(\nu-1)=g(\nu^2)$ gives $\deg g\le n$, hence $\deg w_D\le n$. In particular the values $W_d(D)$ for $1\le d\le n+1$ determine $w_D$ by polynomial interpolation. Finally $\mu(\mu+2)/8=-\frac{3}{32}$ for $\mu=-\frac12$ and for $\mu=-\frac32$.
\end{proof}

\section{The oscillator representation and a supertrace}\label{sec:osc}

Let $A$ be the Weyl algebra over $\Q$ generated by $z$ and $\dd$ with $\dd z-z\dd=1$. It acts on $\Q[z]$, with $z$ acting by multiplication and $\dd$ by differentiation, and the monomials $z^i\dd^j$ ($i,j\ge0$) span $A$. In $A$,
\begin{equation}\label{eq:weyl}
z^i\dd^j z=z^{i+1}\dd^j+j\,z^i\dd^{j-1},\qquad \dd\, z^i\dd^j=z^i\dd^{j+1}+i\,z^{i-1}\dd^j .
\end{equation}

\begin{lemma}\label{lem:faithful}
The operators $z^i\dd^j$, $i,j\ge0$, are linearly independent on $\Q[z]$. In particular the monomials $z^i\dd^j$ form a basis of $A$ and the action of $A$ on $\Q[z]$ is faithful.
\end{lemma}

\begin{proof}
The relation $\dd z=z\dd+1$ rewrites every word as a linear combination of these monomials. Suppose a finite sum $B=\sum_{i,j}a_{ij}z^i\dd^j$ acts by zero. On $z^k$, for all large enough integers $k$, the coefficient of $z^{k+r}$ is
\[
\sum_{i-j=r}a_{ij}\,k(k-1)\cdots(k-j+1).
\]
It vanishes for infinitely many $k$, so it is the zero polynomial in $k$. For each fixed $r$ the falling factorials in this sum have distinct degrees $j$ and leading coefficient $1$; hence every $a_{ij}$ is zero. This also shows that the normal forms are unique, so the action of the abstract Weyl algebra is faithful.
\end{proof}

\begin{lemma}\label{lem:osc}
The assignment $\varphi(E)=-\frac14\dd^2$, $\varphi(F)=z^2$, $\varphi(H)=-\frac12(z\dd+\dd z)$ extends to an algebra homomorphism $\varphi\colon\Ue\to A$. On $\Q[z]$, the maps $v_j\mapsto z^{2j}$ and $v_j\mapsto z^{2j+1}$ are isomorphisms of $\g$-modules $V(-\frac12)\to\Q[z]_{\mathrm{even}}$ and $V(-\frac32)\to\Q[z]_{\mathrm{odd}}$.
\end{lemma}

\begin{proof}
On $\Q[z]$,
\[
Ez^k=-\tfrac14k(k-1)z^{k-2},\qquad Fz^k=z^{k+2},\qquad Hz^k=-(k+\tfrac12)z^k .
\]
Hence $(EF-FE)z^k=-\frac14\bigl((k+2)(k+1)-k(k-1)\bigr)z^k=Hz^k$, and $(HE-EH)z^k=2Ez^k$, $(HF-FH)z^k=-2Fz^k$, because $H$ has eigenvalue $-(k-\frac32)$ on $z^{k-2}$ and $-(k+\frac52)$ on $z^{k+2}$. The relations of $\g$ hold on $\Q[z]$, hence in $A$ by Lemma~\ref{lem:faithful}. For $k=2j$ the formulas read $Ez^{2j}=j(-\frac12-j+1)z^{2j-2}$ and $Hz^{2j}=(-\frac12-2j)z^{2j}$, which is \eqref{eq:verma} at $\mu=-\frac12$; for $k=2j+1$ they read $Ez^{2j+1}=j(-\frac32-j+1)z^{2j-1}$ and $Hz^{2j+1}=(-\frac32-2j)z^{2j+1}$, which is \eqref{eq:verma} at $\mu=-\frac32$.\footnote{This is the oscillator, or metaplectic, representation of $\slt$: the even and odd polynomials are the two irreducible pieces, of lowest weights $\frac12$ and $\frac32$ in the usual convention, which here appear as highest weights $-\frac12$ and $-\frac32$ because $F=z^2$ raises the degree. The Casimir element acts on both pieces by $-\frac{3}{32}$, as the computation $\mu(\mu+2)/8$ at $\mu=-\frac12,-\frac32$ shows.}
\end{proof}

\begin{proposition}\label{prop:scalar}
$\varphi(C_D)=w_D(-\frac{3}{32})\cdot1$ in $A$.
\end{proposition}

\begin{proof}
By Proposition~\ref{prop:lambda} and Lemma~\ref{lem:osc}, $\varphi(C_D)$ acts on $\Q[z]_{\mathrm{even}}$ by $\lambda_D(-\frac12)$ and on $\Q[z]_{\mathrm{odd}}$ by $\lambda_D(-\frac32)$. Both equal $w_D(-\frac{3}{32})$ by Proposition~\ref{prop:wD}, and Lemma~\ref{lem:faithful} gives faithfulness.
\end{proof}

Let $\tau\colon A\to\Q$ be the linear map with
\[
\tau(z^i\dd^j)=\delta_{ij}\,(-1)^j\,\frac{j!}{2^j}.
\]
Since $z^j\dd^j$ multiplies $z^k$ by $k(k-1)\cdots(k-j+1)$ and $\sum_kk(k-1)\cdots(k-j+1)x^k=j!\,x^j(1-x)^{-j-1}$, the number $\tau(a)$ is twice the Abel sum $\lim_{x\to-1^+}\sum_k a_{kk}x^k$ of the supertrace of $a$ on $\Q[z]$, where $az^k=a_{kk}z^k+(\text{other powers})$.\footnote{At $x=-1$ the series $\sum_k k(k-1)\cdots(k-j+1)x^k$ does not converge, which is why $\tau$ is defined directly by the formula and the Abel sum is only mentioned as the origin of the formula. The three properties in Lemma~\ref{lem:tau} are proved from the formula alone.} Only the following properties are used.

\begin{lemma}\label{lem:tau}
\begin{enumerate}
\item[(a)] $\tau(1)=1$.
\item[(b)] $\tau(xa)=-\tau(ax)$ for $x\in\{z,\dd\}$ and $a\in A$.
\item[(c)] For $u_1,\dots,u_m\in\{z,\dd\}$,
\[
\tau(u_1\cdots u_m)=\sum_{M}\ \prod_{\{a<b\}\in M}\tfrac12\,[u_a,u_b],
\]
the sum over the perfect matchings $M$ of $[m]$; here $[z,\dd]=-1$, $[\dd,z]=1$ and $[z,z]=[\dd,\dd]=0$. For odd $m$ both sides vanish.
\end{enumerate}
\end{lemma}

\begin{proof}
(a) is the case $i=j=0$. For (b) it is enough to take $a=z^i\dd^j$. By \eqref{eq:weyl}, $\tau(z^i\dd^jz)=\tau(z^{i+1}\dd^j)+j\,\tau(z^i\dd^{j-1})$. Both terms vanish unless $j=i+1$, and then the sum is $(-1)^j\frac{j!}{2^j}+j(-1)^{j-1}\frac{(j-1)!}{2^{j-1}}=-(-1)^j\frac{j!}{2^j}=-\tau(z\cdot z^i\dd^j)$. In the same way $\tau(\dd z^i\dd^j)=\tau(z^i\dd^{j+1})+i\,\tau(z^{i-1}\dd^j)$ vanishes unless $i=j+1$, and then it equals $(-1)^{j+1}\frac{(j+1)!}{2^{j+1}}+(j+1)(-1)^j\frac{j!}{2^j}=(-1)^j\frac{(j+1)!}{2^{j+1}}=-\tau(z^i\dd^j\cdot\dd)$.

(c) Induction on $m$. For $m=0$ use (a); for $m=1$, $\tau(z)=\tau(\dd)=0$. For $m\ge2$, as the $[u_1,u_s]$ are scalars,
\[
u_1u_2\cdots u_m=\sum_{s=2}^m[u_1,u_s]\,u_2\cdots\widehat{u_s}\cdots u_m+u_2\cdots u_mu_1 .
\]
Applying $\tau$ and (b) to the last term gives
\[
\tau(u_1\cdots u_m)=\tfrac12\sum_{s=2}^m[u_1,u_s]\,\tau(u_2\cdots\widehat{u_s}\cdots u_m),
\]
which is the expansion of the right side of (c) according to the partner $s$ of $1$.
\end{proof}

\begin{corollary}\label{cor:tau}
$w_D(-\frac{3}{32})=\tau(\varphi(C_D))$.
\end{corollary}

\section{Expansion over matchings}\label{sec:matching}

\subsection{Matchings}
Let $X$ be a finite totally ordered set and $P,M$ perfect matchings of $X$. Two edges $\{a<b\}$ and $\{a'<b'\}$ cross if $a<a'<b<b'$ or $a'<a<b'<b$, and $\cro(P)$ is the number of crossing pairs of edges of $P$. The multigraph $P\cup M$ is a disjoint union of cycles alternating between edges of $P$ and of $M$ (a common edge gives a cycle of length $2$); $\cyc(P,M)$ is their number. Orient every cycle; an edge oriented from a larger to a smaller element is a descent, and $\des(P,M)$ is the number of descents. Reversing a cycle of length $2r$ turns $\delta$ descents into $2r-\delta$, so the parity of $\des(P,M)$ does not depend on the orientations. Put
\[
L_P(N)=\sum_M(-1)^{\cro(M)}N^{\cyc(P,M)}\in\Z[N],
\]
the sum over all perfect matchings $M$ of $X$.

\begin{lemma}\label{lem:sign}
$\cro(P)+\cro(M)+\cyc(P,M)+\des(P,M)$ is even.
\end{lemma}

\begin{proof}
Let $Q(P,M)$ be the sum. Let $x<y$ be consecutive in $X$, and exchange their positions in the order, keeping $P$ and $M$. If $\{x,y\}\notin P$, let $e\ni x$ and $f\ni y$ be the edges of $P$; the exchange moves $y$ to the other side of the endpoint $x$ of $e$ and keeps the side of the other endpoint of $f$, so it changes whether $e$ and $f$ cross, and no other pair. So $\cro(P)$ changes by $1-[\{x,y\}\in P]$ modulo $2$, and likewise $\cro(M)$. $\cyc(P,M)$ does not change. A descent status changes only on edges joining $x$ and $y$, so $\des(P,M)$ changes by $[\{x,y\}\in P]+[\{x,y\}\in M]$. The total change is $2$. Any two total orders of $X$ are related by exchanges of consecutive elements, so $Q$ modulo $2$ is the same for all orders. Choose an order in which every cycle $x_1x_2\cdots x_{2r}$ (with $x_1x_2\in P$) occupies consecutive positions in this order. Then $P$ and $M$ have no crossings, and each cycle, oriented as listed, has exactly one descent, from $x_{2r}$ to $x_1$. Hence $Q\equiv2\cyc(P,M)\equiv0$.
\end{proof}

\subsection{The bosonic expansion}
For a perfect matching $P$ of $X$ with $k$ edges let
\[
X_P=\sum_{o}\operatorname{sgn}(o)\prod_{x\in X}^{\longrightarrow}u^o_x\in A,
\]
the sum over the $2^k$ orientations $o$ of the edges of $P$, where $u^o_x=z$ if $x$ is the tail of its edge and $u^o_x=\dd$ if it is the head, the product is taken in the order of $X$, and $\operatorname{sgn}(o)=(-1)^{\text{number of descents of }o}$.

\begin{proposition}\label{prop:boson}
$\tau(X_P)=2^{-k}(-1)^{\cro(P)}L_P(-2)$.
\end{proposition}

\begin{proof}
By Lemma~\ref{lem:tau}(c),
\[
\tau(X_P)=\sum_o\sum_M\operatorname{sgn}(o)\prod_{\{a<b\}\in M}\tfrac12[u^o_a,u^o_b].
\]
A term vanishes unless every edge of $M$ joins a head and a tail of $o$. Orient such an edge from the head to the tail; then $\frac12[u^o_a,u^o_b]$ is $\frac12$ if it is oriented from $a$ to $b$ ($u^o_a=\dd$, $u^o_b=z$) and $-\frac12$ if it is a descent. As $\operatorname{sgn}(o)$ is $-1$ to the number of descents among the edges of $P$, the term equals $2^{-k}(-1)^{\des(P,M)}$. Every element of $X$ is then entered by exactly one oriented edge and left by exactly one, so the cycles of $P\cup M$ are directed cycles. Conversely, each of the $2^{\cyc(P,M)}$ orientations of the cycles of $P\cup M$ as directed cycles restricts on $P$ to an orientation $o$ for which every edge of $M$ joins a head to a tail. Hence
\begin{align*}
\tau(X_P)&=2^{-k}\sum_M2^{\cyc(P,M)}(-1)^{\des(P,M)}\\
&=2^{-k}(-1)^{\cro(P)}\sum_M(-1)^{\cro(M)}(-2)^{\cyc(P,M)}
\end{align*}
by Lemma~\ref{lem:sign}.
\end{proof}

\subsection{Legs}
Let $D$ be a chord diagram with $n$ chords. Every point $p\in[2n]$ carries two legs $(p,1)<(p,2)$, and the set $X_D$ of the $4n$ legs is ordered lexicographically. For a chord $c=\{p<q\}$ and $\beta\in\{0,1\}$ let $P_c^\beta$ be the pair of edges
\[
P_c^0=\{(p,1)(q,1),\ (p,2)(q,2)\},\qquad P_c^1=\{(p,1)(q,2),\ (p,2)(q,1)\},
\]
and for $\beta\colon\{\text{chords}\}\to\{0,1\}$ let $P_\beta=\bigcup_cP_c^{\beta(c)}$, a perfect matching of $X_D$. Let $\pi=z\otimes\dd-\dd\otimes z\in A\otimes A$. Writing $\pi=\sum_r\varepsilon_r\,x_r\otimes y_r$, put $\pi_{13}\pi_{24}=\sum_{r,s}\varepsilon_r\varepsilon_s\,x_rx_s\otimes y_ry_s$ and $\pi_{14}\pi_{23}=\sum_{r,s}\varepsilon_r\varepsilon_s\,x_rx_s\otimes y_sy_r$.

\begin{lemma}\label{lem:pi}
$(\varphi\otimes\varphi)(t)=-\frac{1}{32}\bigl(\pi_{13}\pi_{24}+\pi_{14}\pi_{23}\bigr)$.
\end{lemma}

\begin{proof}
Both sides equal $-\frac1{32}\bigl(2z^2\otimes\dd^2+2\dd^2\otimes z^2-(z\dd+\dd z)\otimes(z\dd+\dd z)\bigr)$. On the left, $\frac14\varphi(E)\otimes\varphi(F)+\frac14\varphi(F)\otimes\varphi(E)=-\frac1{16}(\dd^2\otimes z^2+z^2\otimes\dd^2)$ and $\frac18\varphi(H)\otimes\varphi(H)=\frac1{32}(z\dd+\dd z)^{\otimes2}$. On the right, the terms with $r=s$ give $2z^2\otimes\dd^2+2\dd^2\otimes z^2$ and the terms with $r\ne s$ give $-z\dd\otimes(\dd z+z\dd)-\dd z\otimes(z\dd+\dd z)$.
\end{proof}

\begin{corollary}\label{cor:legs}
$\varphi(C_D)=\bigl(-\frac{1}{32}\bigr)^n\sum_\beta X_{P_\beta}$, the sum over all $\beta\colon\{\text{chords}\}\to\{0,1\}$.
\end{corollary}

\begin{proof}
$\varphi(C_D)$ is the product, in the order of the points, of the factors of $(\varphi\otimes\varphi)(t)^{\otimes n}$ placed at the endpoints of the chords. Put the two letters of each factor $x_rx_s$ or $y_ry_s$ on the two legs of its point, in this order. By Lemma~\ref{lem:pi}, a chord contributes $-\frac1{32}$ times the sum of $\pi$ placed on the two edges of $P_c^0$ and $\pi$ placed on the two edges of $P_c^1$. A copy of $\pi$ on an edge $\{a<b\}$ contributes $z$ at $a$ and $\dd$ at $b$ with sign $+1$, or $\dd$ at $a$ and $z$ at $b$ with sign $-1$; these are the two orientations of the edge with their signs in $X_{P_\beta}$.
\end{proof}

\section{Clifford algebras}\label{sec:clifford}

For $N\ge1$ let $Cl_N$ be the $\Q$-algebra generated by $e_1,\dots,e_N$ with $e_ie_j+e_je_i=2\delta_{ij}$. For $I=\{i_1<\dots<i_r\}\subseteq[N]$ put $e_I=e_{i_1}\cdots e_{i_r}$; these form a basis, and $e_Ie_J=\pm e_{I\triangle J}$. Let $\tr\colon Cl_N\to\Q$ be linear with $\tr(e_I)=\delta_{I\emptyset}$. Then $\tr(ab)=\tr(ba)$: for basis elements both sides vanish unless $I=J$.\footnote{For $I=J$ both products are $e_Ie_I=\pm1$ with the same sign, since $e_Ie_I$ and $e_Ie_I$ are the same word. For $I\ne J$ the products are $\pm e_{I\triangle J}$ with $I\triangle J\ne\emptyset$, so both traces vanish. The trace is $2^{-N}$ times the trace of the left regular representation.}

\begin{lemma}\label{lem:cliffwick}
For $i_1,\dots,i_m\in[N]$,
\[
\tr(e_{i_1}\cdots e_{i_m})=\sum_M(-1)^{\cro(M)}\prod_{\{a,b\}\in M}\delta_{i_ai_b},
\]
the sum over the perfect matchings $M$ of $[m]$.
\end{lemma}

\begin{proof}
For odd $m$ the product is $\pm e_I$ with $|I|$ odd, so both sides vanish. For even $m\ge2$, moving $e_{i_1}$ to the right end gives
\[
e_{i_1}e_{i_2}\cdots e_{i_m}=\sum_{s=2}^m(-1)^{s}\,2\delta_{i_1i_s}\,e_{i_2}\cdots\widehat{e_{i_s}}\cdots e_{i_m}-e_{i_2}\cdots e_{i_m}e_{i_1}.
\]
Taking traces and using $\tr(e_{i_2}\cdots e_{i_m}e_{i_1})=\tr(e_{i_1}\cdots e_{i_m})$,
\[
\tr(e_{i_1}\cdots e_{i_m})=\sum_{s=2}^m(-1)^s\delta_{i_1i_s}\tr(e_{i_2}\cdots\widehat{e_{i_s}}\cdots e_{i_m}).
\]
The edge $\{1,s\}$ crosses exactly the edges with one endpoint in $\{2,\dots,s-1\}$; the elements of this set matched among themselves come in pairs, so the number of such edges is congruent to $s-2$ modulo $2$. By induction both sides satisfy the same recursion.
\end{proof}

\begin{corollary}\label{cor:colour}
For a perfect matching $P$ of a totally ordered set $X=\{x_1<\dots<x_{2k}\}$ and $N\ge1$,
\[
\sum_\kappa\tr(e_{\kappa(x_1)}\cdots e_{\kappa(x_{2k})})=L_P(N),
\]
the sum over the maps $\kappa\colon X\to[N]$ that are constant on every edge of $P$.
\end{corollary}

\begin{proof}
By Lemma~\ref{lem:cliffwick} the left side is $\sum_M(-1)^{\cro(M)}$ times the number of maps $X\to[N]$ constant on the edges of $P$ and of $M$, which is $N^{\cyc(P,M)}$.
\end{proof}

For $I=\{i<j\}\in\binomset{[N]}{2}$ write $b_I=e_ie_j$. Then
\begin{equation}\label{eq:bivector}
b_Ib_J=(-1)^{|I\cap J|}\,b_Jb_I,\qquad b_I^2=-1 .
\end{equation}

\begin{proposition}\label{prop:clifford}
Let $D$ be a chord diagram with $n$ chords and $\Gamma=\Gamma(D)$. Then
\[
\sum_\beta(-1)^{\cro(P_\beta)}L_{P_\beta}(N)=4^n\,S_\Gamma(N)
\]
as polynomials in $N$.
\end{proposition}

\begin{proof}
Both sides are polynomials (Theorem~\ref{thm:main}(a) is proved in Section~\ref{sec:proof} on its own), so it is enough to take $N\ge1$.

The two edges of $P_c^\beta$ cross if and only if $\beta=0$. If chords $c$ and $c'$ cross, every edge of $P_c^\beta$ crosses every edge of $P_{c'}^{\beta'}$; otherwise none does. Hence
\[
(-1)^{\cro(P_\beta)}=(-1)^n\prod_c(-1)^{\beta(c)}.
\]

A map $\kappa$ constant on the edges of $P_\beta$ is given by a pair $(i_c,j_c)\in[N]^2$ for every chord $c=\{p<q\}$: $i_c$ on the edge through $(p,1)$ and $j_c$ on the edge through $(p,2)$. The legs at $p$ then read $e_{i_c}e_{j_c}$, and the legs at $q$ read $e_{i_c}e_{j_c}$ if $\beta(c)=0$ and $e_{j_c}e_{i_c}$ if $\beta(c)=1$. By Corollary~\ref{cor:colour}, summing over $\beta(c)$ with the sign $(-1)^{\beta(c)}$ replaces the factor at $q$ by $e_{i_c}e_{j_c}-e_{j_c}e_{i_c}$, which is $0$ if $i_c=j_c$ and $2e_{i_c}e_{j_c}$ otherwise. Exchanging $i_c$ and $j_c$ changes the sign of both factors of $c$, so every $2$-element set $s(c)=\{i_c,j_c\}$ arises twice with the same value. Therefore
\[
\sum_\beta(-1)^{\cro(P_\beta)}L_{P_\beta}(N)=(-1)^n4^n\sum_{s}\tr(B_s),\qquad B_s=\prod_{p=1}^{2n}b_{s(c_p)},
\]
the sum over $s\colon\{\text{chords}\}\to\binomset{[N]}{2}$, where $c_p$ is the chord through $p$.

We show $\tr(B_s)=(-1)^n\prod_{cc'\in E(\Gamma)}(-1)^{|s(c)\cap s(c')|}$ by induction on $n$. Let $c=\{p<q\}$. Move the factor at $p$ to the right until it is next to the factor at $q$. By \eqref{eq:bivector} it passes the factors at the points strictly between $p$ and $q$; a chord with both endpoints there contributes the square of a sign, and the chords with one endpoint there are the neighbours of $c$ in $\Gamma$. Then $b_{s(c)}^2=-1$, and the remaining word is $B_{s'}$ for the diagram $D\setminus c$, whose intersection graph is $\Gamma-c$, with $s'=s|_{\Gamma-c}$. Summing over $s$ gives $\sum_s\tr(B_s)=(-1)^nS_\Gamma(N)$.
\end{proof}

\section{Proof of Theorem~\ref{thm:main}}\label{sec:proof}

\begin{proof}[Proof of (a)]
Group the maps $p\colon V\to\binomset{[N]}{2}$ by $U=\bigcup_xp(x)$. Relabelling $U$ in increasing order by $[|U|]$ preserves all intersection sizes, so the maps with $|U|=k$ contribute $\binom Nk T_k(G)$; and $T_k(G)=0$ for $k>2n$.
\end{proof}

\begin{proof}[Proof of (c)]
By Corollaries~\ref{cor:tau} and~\ref{cor:legs}, Proposition~\ref{prop:boson} with $k=2n$, and Proposition~\ref{prop:clifford},
\begin{align*}
w_D(-\tfrac{3}{32})&=\Bigl(-\frac{1}{32}\Bigr)^n\sum_\beta\tau(X_{P_\beta})=\Bigl(-\frac{1}{32}\Bigr)^n4^{-n}\sum_\beta(-1)^{\cro(P_\beta)}L_{P_\beta}(-2)\\
&=\frac{S_\Gamma(-2)}{(-32)^n}=\xi(\Gamma).\qedhere
\end{align*}
\end{proof}

For a graph $G$ and distinct vertices $a,b$ let $G'_{ab}$ be $G$ with the adjacency of $a$ and $b$ reversed, and $\widetilde G_{ab}$ be $G$ with the adjacency of $a$ and $y$ reversed for every neighbour $y\ne a$ of $b$. The four-term relation for a function $f$ on graphs is
\[
f(G)-f(G'_{ab})=f(\widetilde G_{ab})-f(\widetilde G'_{ab}),\qquad \widetilde G'_{ab}=(\widetilde G_{ab})'_{ab},
\]
as in Figure~\ref{fig:fourterm}.\footnote{This is Lando's relation \cite{Lan00}. For intersection graphs it is the image of the four-term relation of chord diagrams: if the chords $a$ and $b$ share an endpoint region, then moving one endpoint of $a$ past the endpoint of $b$ reverses the adjacency of $a$ with $b$, and moving it past the other endpoint of $b$ reverses the adjacency of $a$ with every chord that crosses $b$ and not $a$, which is the passage to $\widetilde G_{ab}$. So a function on graphs satisfying the relation gives a weight system.}

\begin{figure}[ht]
\centering
\includegraphics[width=\textwidth]{figures/fig_fourterm.png}
\caption{The four graphs of a four-term relation. Here $b$ has the neighbours $a,y_1,y_2$; passing from $G$ to $\widetilde G_{ab}$ removes the edge $ay_1$ (dashed) and adds the edge $ay_2$.}\label{fig:fourterm}
\end{figure}

\begin{proposition}\label{prop:fourT}
For every $N\in\N$, $G\mapsto S_G(N)$ satisfies the four-term relation.
\end{proposition}

\begin{proof}
For $p\colon V\to\binomset{[N]}{2}$ write
\[
\epsilon(G,p)=\prod_{xy\in E(G)}(-1)^{|p(x)\cap p(y)|}.
\]
Since $\epsilon(G'_{ab},p)=\epsilon(G,p)(-1)^{|p(a)\cap p(b)|}$ and $|p(a)\cap p(b)|\in\{0,1,2\}$,
\[
S_G(N)-S_{G'_{ab}}(N)=2\sum_{p\in\Pi}\epsilon(G,p),\qquad \Pi=\{p:|p(a)\cap p(b)|=1\}.
\]
The construction $\widetilde G_{ab}$ does not change the adjacency of $a$ and $b$, so the same holds for $\widetilde G_{ab}$. On $\Pi$ define $\iota(p)$ by $\iota(p)(b)=p(a)\triangle p(b)$ and $\iota(p)(x)=p(x)$ for $x\ne b$. As $|p(a)\cap p(b)|=1$, $\iota(p)(b)$ is a $2$-element set meeting $p(a)$ in one element, and $\iota$ is an involution of $\Pi$.\footnote{If $p(a)=\{u,v\}$ and $p(b)=\{v,w\}$ with $u,v,w$ distinct, then $\iota(p)(b)=\{u,w\}$, and applying $\iota$ again gives $\{u,v\}\triangle\{u,w\}=\{v,w\}=p(b)$.} Let $Y$ be the set of neighbours $y\ne a$ of $b$ in $G$, which is also the set of such neighbours in $\widetilde G_{ab}$. For $y\in Y$, $|(p(a)\triangle p(b))\cap p(y)|\equiv|p(a)\cap p(y)|+|p(b)\cap p(y)|$ modulo $2$, and the edge $ab$, if present, keeps its sign because $|p(a)\cap\iota(p)(b)|=1$. Hence
\begin{gather*}
\epsilon(G,\iota(p))=\epsilon(G,p)\prod_{y\in Y}(-1)^{|p(a)\cap p(y)|},\\
\epsilon(\widetilde G_{ab},q)=\epsilon(G,q)\prod_{y\in Y}(-1)^{|q(a)\cap q(y)|},
\end{gather*}
the second for every $q$, since $\widetilde G_{ab}$ differs from $G$ exactly in the pairs $ay$, $y\in Y$. With $q=\iota(p)$, which agrees with $p$ at $a$ and on $Y$, this gives $\epsilon(\widetilde G_{ab},\iota(p))=\epsilon(G,p)$. Therefore $\sum_{p\in\Pi}\epsilon(\widetilde G_{ab},p)=\sum_{p\in\Pi}\epsilon(G,p)$, and the two differences agree.
\end{proof}

\begin{proof}[Proof of (b)]
$\xi$ depends only on the isomorphism class of $G$. The four graphs of a four-term relation have the same vertex set, so by Proposition~\ref{prop:fourT} and (a) the four-term relation holds for the polynomials $S_G$, hence at $N=-2$, and after division by $(-32)^n$ for $\xi$.
\end{proof}

\begin{remark}\label{rem:Tk}
Proposition~\ref{prop:fourT} applied to the sets $\Pi$ restricted to the maps with $\bigcup_xp(x)=[k]$, which $\iota$ preserves, shows that every $T_k$ satisfies the four-term relation. By the proof of Proposition~\ref{prop:clifford}, for $N\ge1$ the $4$-invariant $S_\Gamma(N)$ is $(-4)^{-n}$ times the Clifford trace of the element obtained by placing the tensor $\sum_{i\ne j}e_ie_j\otimes 2e_ie_j$ on the chords as in the definition of $C_D$, and Theorem~\ref{thm:main}(c) identifies $w_D(-\frac{3}{32})$ with $(-32)^{-n}$ times its value at $N=-2$.
\end{remark}

\begin{remark}\label{rem:scope}
The formula for $\xi$ is a finite formula for every finite graph, with no bound on the number of vertices, since $T_k(G)=0$ for $k>2|V(G)|$. The involution of Proposition~\ref{prop:fourT} is defined on every assignment with $|p(a)\cap p(b)|=1$, so the proof of (b) does not rest on a table of graphs. The identities of Sections~\ref{sec:verma} to~\ref{sec:clifford} hold for every chord diagram. The evaluation at $N=-2$ in the proof of (c) is substitution into a polynomial; no limit or analytic continuation is involved. The finite-dimensional normalised traces $W_d(D)$ are the starting definition, and the Verma module argument proves that they interpolate to a unique polynomial before its value at the oscillator representation is taken. The theorem says nothing about the unspecialised weight system, which does not extend to graphs by \cite{FKPU}.
\end{remark}

\section*{Declarations}
Funding: none. Conflict of interest: none.

\begin{thebibliography}{10}
\raggedright

\bibitem{Kon93}
M.~Kontsevich, Vassiliev's knot invariants, in: I. M. Gel'fand Seminar, Adv. Soviet Math. 16, Part 2, Amer. Math. Soc., Providence, RI, 1993, pp. 137--150. \href{https://doi.org/10.1090/advsov/016.2/04}{doi:10.1090/advsov/016.2/04}.

\bibitem{BN95}
D.~Bar-Natan, On the Vassiliev knot invariants, Topology 34 (1995), no. 2, 423--472. \href{https://doi.org/10.1016/0040-9383(95)93237-2}{doi:10.1016/0040-9383(95)93237-2}.

\bibitem{CDM12}
S.~Chmutov, S.~Duzhin, J.~Mostovoy, Introduction to Vassiliev Knot Invariants, Cambridge University Press, Cambridge, 2012. \href{https://doi.org/10.1017/CBO9781139107846}{doi:10.1017/CBO9781139107846}.

\bibitem{ChL07}
S.~Chmutov, S.~Lando, Mutant knots and intersection graphs, Algebr. Geom. Topol. 7 (2007), no. 3, 1579--1598. \href{https://doi.org/10.2140/agt.2007.7.1579}{doi:10.2140/agt.2007.7.1579}.

\bibitem{Lan00}
S.~K. Lando, On a Hopf algebra in graph theory, J. Combin. Theory Ser. B 80 (2000), no. 1, 104--121. \href{https://doi.org/10.1006/jctb.2000.1973}{doi:10.1006/jctb.2000.1973}.

\bibitem{KL22}
M.~E. Kazarian, S.~K. Lando, Weight systems and invariants of graphs and embedded graphs, Russian Math. Surveys 77 (2022), no. 5, 893--942. \href{https://doi.org/10.4213/rm10054e}{doi:10.4213/rm10054e}.

\bibitem{Kra21}
E.~Krasilnikov, An extension of the $\mathfrak{sl}_2$ weight system to graphs with $n\le8$ vertices, Arnold Math. J. 7 (2021), 609--618. \href{https://doi.org/10.1007/s40598-021-00187-7}{doi:10.1007/s40598-021-00187-7}.

\bibitem{FKPU}
D.~Fomichev, M.~Karev, F.~Pavutnitskiy, S.~Usanov, $\mathfrak{sl}(2)$-weight system does not extend to a graph 4-invariant, arXiv:2607.24217 (2026).

\bibitem{FK25}
D.~Fomichev, M.~Karev, The $\mathfrak{sl}(2)$-weight system at $c=3/8$ for graphs, European J. Combin. 127 (2025), 104160. \href{https://doi.org/10.1016/j.ejc.2025.104160}{doi:10.1016/j.ejc.2025.104160}.

\bibitem{YDJ26}
Q.~Yan, Q.~Deng, X.~Jin, Proof of a conjecture of Fomichev and Karev, European J. Combin. 135 (2026), 104365. \href{https://doi.org/10.1016/j.ejc.2026.104365}{doi:10.1016/j.ejc.2026.104365}.

\bibitem{ChV97}
S.~V. Chmutov, A.~N. Varchenko, Remarks on the Vassiliev knot invariants coming from $\mathfrak{sl}_2$, Topology 36 (1997), no. 1, 153--178. \href{https://doi.org/10.1016/0040-9383(95)00071-2}{doi:10.1016/0040-9383(95)00071-2}.

\bibitem{Hum08}
J.~E. Humphreys, Representations of Semisimple Lie Algebras in the BGG Category $\mathcal{O}$, Grad. Stud. Math. 94, Amer. Math. Soc., Providence, RI, 2008. \href{https://doi.org/10.1090/gsm/094}{doi:10.1090/gsm/094}.

\bibitem{Bha22a}
D.~Bhattacharjee, Establishing equivariant class $[O]$ for hyperbolic groups, Asian Res. J. Math. 18 (2022), no. 11, 362--369. \href{https://doi.org/10.9734/arjom/2022/v18i11615}{doi:10.9734/arjom/2022/v18i11615}.

\bibitem{Bha22b}
D.~Bhattacharjee, Non-commutativity over canonical suspension $\eta$ for genus $g\ge1$ in hypercomplex structures for potential $\rho\phi$, Asian Res. J. Math. 18 (2022), no. 11, 332--341. \href{https://doi.org/10.9734/arjom/2022/v18i11607}{doi:10.9734/arjom/2022/v18i11607}.

\bibitem{BB26}
D.~Bhattacharjee, U.~Bhattacharya, Iterated nerve complexes for $k$-fold sensor coverage, Preprints.org (2026). \href{https://doi.org/10.20944/preprints202606.1351.v2}{doi:10.20944/preprints202606.1351.v2}.

\end{thebibliography}

\end{document}