Satellite ruling polynomials, DGA representations, and the colored HOMFLY-PT polynomial (Q2179904)

A Legendrian knot in a contact manifold \((M,\xi)\) is an embedding \(S^1\to M\) so that the tangent space of the image is contained in \(\xi\). If \(\Sigma\) is a connected Seifert surface for a Legendrian knot \(K\), then the Thurston-Bennequin invariant \(\mathsf{tb}(K)\) of \(K\) is the algebraic intersection number of the push off of \(K\) along a vector field transverse to \(\xi\) with \(\Sigma\), and the rotation number \(\mathsf{rot}(K)\) of \(K\) is the number of full turns the positive tangent vector of \(K\) makes compared to an oriented framing of \(\xi\vert_{\Sigma}\) as one moves along the oriented curve. If \(L\) is a Legendrian representative of an integrally null-homologous knot \(K\) in a tight contact 3-manifold \((Y,\xi)\), then the well-known Bennequin-Eliashberg inequality is \(\mathsf{tb}(L)+\mathsf{rot}(L)\le 2g(K)-1\), where \(g(K)\) is the genus of \(K\). If \(\alpha=dz-ydx\) is the standard contact structure with the front \((xz)\)-projection and the Lagrangian \((xy)\)-projection, then \(J^1\mathbb{R}^1\cong\mathbb{R}^3\) and \(J^1\mathbb{S}^1\cong S^1\times\mathbb{R}^2\) are the \(1\)-jet spaces coordinatized as \((x,y,z)\). If \(K\subset J^1\mathbb{R}^1\) or \(J^1\mathbb{S}^1\) is a Legendrian link equipped with some collection of basepoints, then \((\mathscr{A}(K),\partial)\) is the fully non-commutative version of the Chekanov-Eliashberg DGA, also called the Legendrian contact homology DGA, defined via the \((xy)\)-projection of \(K\). When \(K\) is connected and has multiple components \(K=\bigsqcup\limits_{i=1}^cK_i\), then the rotation number \(\mathsf{rot}(K)\) of \(K\) is the greatest common divisor of \(\mathsf{rot}(K_i)\). For a non-negative integer \(m>0\) and a Legendrian \(K\) in \(J^1\mathbb{R}^1\) or \(J^1\mathbb{S}^1\), a \(\mathbb{Z}/m\)-valued function \(\mu\) on \(K\) that is locally constant except at cusp points, where the value at the upper branch of the cusp is \(1\) larger than at the lower branch, is called \(\mathbb{Z}/m\)-valued Maslov potential. A connected Legendrian \(K\) has a \(\mathbb{Z}/m\)-valued Maslov potential if and only if \(m|2r(K)\). If \(K\) is a Legendrian link in \(J^1\mathbb{R}^1\) or \(J^1\mathbb{S}^1\) equipped with a \(\mathbb{Z}/m\)-valued Maslov potential \(\mu\) with \(m|2r(K)\), then the \(m\)-graded ruling polynomial of \(K\) is defined as \(R^m_K(z)=\sum\limits_{\rho}z^{j(\rho)}\), where the sum is over all \(m\)-graded normal rulings of \(K\) and \(j(p)=\#(\text{switches})-\#(\text{rightcusps})\), and an \(m\)-graded normal ruling of \(K\) is a combinatorial structure associated to the front projection of \(K\). In this paper, the authors establish relationships between two classes of invariants of Legendrian knots, representation numbers of the Chekanov-Eliashberg DGA and satellite ruling polynomials in \(\mathbb{R}^3\). If the Legendrian knot \(K\subset J^1\mathbb{R}\) has rotation number \(\mathsf{rot}(K)=0\), \(m\ge 0\) is a non-negative integer, and \(\mathbb{F}_q\) is a finite field of order \(q\) (when \(m\) is odd, \(\mathrm{char}(\mathbb{F}_q)=2\)), then the authors introduce Legendrian isotopy invariant \(m\)-graded representation numbers denoted \(\mathrm{Rep}_m(K,(V, d), B)\) that count representations of \((\mathcal{A}(K),\partial)\) on a differential graded vector space \((V,d)\) over \(\mathbb{F}_q\) where certain distinguished generators, \(t_i\), associated to basepoints on \(K\) have their images restricted by a choice of subset \(B\subset\mathrm{GL}(V)\). Also there are considered reduced representation numbers, \(\widetilde{\mathrm{Rep}}_m(K,(V,d),B)\), that are renormalized so the unknot takes the value \(1\), and a total \(n\)-dimensional representation number, \(\mathrm{Rep}_m(K,\mathbb{F}^n_q)\), that counts all \(m\)-graded representations to \(\mathbb{F}^n_q\) without restriction on the image of the \(t_i\). The authors prove that if \(\beta\subset J^1S^1\) is an \(n\)-stranded positive permutation braid, then \[ R^m_{S(K,\beta}(z)_{z=q^{1/2}-q^{-1/2}}=q^{-\lambda_m(\beta)/2}\left(q^{1/2}-q^{-1/2}\right)^{-n}\sum\limits_d\widetilde{\mathrm{Rep}}_m(K,(V_\beta,d),B^m_\beta), \] where \(V_\beta\) is a graded vector space over \(\mathbb{F}_q\) associated to \(\beta\), \(B_\beta\subset\mathrm{GL}(V_\beta)\) is the path subset of \(\beta\), \(\lambda_m(\beta)\) is a certain signed count of the crossings of \(\beta\), and the summation is over all strictly upper triangular differentials, \(d:V_\beta\to V_\beta\), with \(\mathrm{deg}(d)=\pm 1\). Next, the authors define the notion of the \(n\)-colored \(m\)-graded ruling polynomial. For any fixed \(L\subset J^1S^1\) the satellite ruling polynomial \(R^m_{S(K,L)}\) is determined by the Chekanov-Eliashberg DGA of \(K\) and the \(n\)-colored ruling polynomial is a specific linear combination of satellite ruling polynomials. It is so named because of a strong formal relation with the \(n\)-colored HOMFLY-PT polynomial \(P_{n,K}(a,q)\). The authors also obtain a converse-type formula for representation numbers in terms of ruling polynomials by stating that if \(m\ne 1\), then the total \(n\)-dimensional \(m\)-graded representation number satisfies \[ \mathrm{Rep}_m(K,\mathbb{F}^n_q)=R^m_{n,K}(q), \] where \(R^m_{n,K}(q)\) denotes the \(n\)-colored \(m\)-graded ruling polynomial. Finally, the authors show that \(\mathsf{tb}(L)+\mathsf{rot}(L)\le \frac{1}{n}\mathrm{deg}_a\widehat{P}_{n,K}(a,q)\), where \(\widehat{P}_{n,K}\) is the framing independent version of \({P}_{n,K}\).