--- title: 'Braid Charts: Combinatorial Encodings' url: https://www.emergentmind.com/topics/braid-charts type: topic --- # Braid Charts: Combinatorial Encodings Braid charts are combinatorial descriptions of braid monodromy and branched coverings. In the classical setting, a chart of degree \(d\) is a finite oriented labeled graph on a compact oriented surface, most often a disk or sphere, whose local vertex types model simple branch points, commutation, and the braid relation; intersection words obtained from transverse loops recover the monodromy of a 2-dimensional braid. In this sense charts convert the topological problem of describing branched coverings into a combinatorial one in terms of finite labeled graphs, and the formalism extends to surface-links, to 3-dimensional braids via curtains, and, in later algebraic work, to affine and toric chart systems associated with braid varieties and Bott–Samelson varieties [1312.5388, 1206.4744, 2606.05823]. ## 1. Classical definitions and local models Let \(\Sigma^2\) be a compact, oriented surface and fix an integer \(d\ge 2\). A chart of degree \(d\) is a finite, oriented, labeled graph \(\Gamma\subset \Sigma^2\) such that no vertex lies on \(\partial \Sigma^2\), every edge is oriented and labeled in \(\{1,2,\dots,d-1\}\), and every vertex is one of three types: a black vertex of valency \(1\), a crossing of valency \(4\), or a white vertex of valency \(6\) [1312.5388]. | Vertex type | Valency | Local meaning | |---|---:|---| | Black vertex | 1 | Simple branch point | | Crossing | 4 | Two transversely intersecting edges with labels \(i,j\), \(|i-j|>1\) | | White vertex | 6 | Local braid relation for labels \(i,i+1\) | The crossing models commutation when \(|i-j|>1\), and the white vertex models the braid relation \[ \sigma_i\sigma_{i+1}\sigma_i=\sigma_{i+1}\sigma_i\sigma_{i+1}. \] When \(\Sigma^2\) has nonempty boundary, one may allow \(1\)-valent vertices on \(\partial \Sigma^2\); these are external boundary vertices, the “ends” of free edges [1312.5388]. Several closely related chart conventions appear in the literature. On a surface diagram \(D\) of an oriented surface-knot \(F\subset \mathbb R^4\), a chart of degree \(N\) is a finite labeled oriented graph \(\Gamma\subset D\) whose vertices have degree \(1,2,4\), or \(6\); the degree-\(2\) vertices occur where \(\Gamma\) crosses a sheet-double curve of \(D\) [1503.00404]. In another formulation used for knotted surfaces in \(\mathbb R^4\), a degree-\(n\) braid chart \(\Gamma\subset D^2\) is an embedded oriented graph whose edges are labeled by \[ \ell(e)\in\{\pm 1,\dots,\pm(n-1)\}, \] with black vertices interpreted as sources or sinks, white vertices as Reidemeister III events, and \(4\)-valent crossings recording the order of braid generators [2510.04248]. This suggests that “b braid chart” is a stable local-combinatorial concept, while the global ambient space and the precise sign convention vary with the application. ## 2. Encoding braid monodromy Suppose \(S\subset D^2\times B^2\) is a \(2\)-dimensional braid of degree \(d\), so that \(\operatorname{pr}_2|_S:S\to B^2\) is a simple branched covering of degree \(d\) with \(\partial S=X_d\times \partial B^2\). For a chart \(\Gamma\subset B^2\), let \(\Delta(\Gamma)\subset \operatorname{int} B^2\) be the set of black vertices, choose \(q_0\in \partial B^2\), and let \[ \alpha:([0,1],\{0,1\})\to (B^2\setminus \Delta(\Gamma),q_0) \] be a loop transverse to \(\Gamma\). If \(\alpha\) meets edges \(e_1,\dots,e_k\) in order, then each intersection contributes \(\sigma_{\ell(e_j)}^{\epsilon_j}\), where \(\epsilon_j=\pm 1\) depends on whether \(\alpha\) crosses the edge in the positive or negative direction relative to the edge’s arrow. The resulting intersection word is \[ \operatorname{Intersection\ Word}(\alpha)= \sigma_{\ell(e_1)}^{\epsilon_1}\sigma_{\ell(e_2)}^{\epsilon_2}\cdots \sigma_{\ell(e_k)}^{\epsilon_k}\in B_d, \] and the associated homomorphism \[ \rho_\Gamma:\pi_1(B^2\setminus \Delta(\Gamma),q_0)\to B_d,\qquad [\alpha]\mapsto \operatorname{Intersection\ Word}(\alpha) \] agrees with the braid monodromy \(\rho_S\) of \(S\) [1312.5388]. The correspondence is complete: for every \(2\)-dimensional braid \(S\subset D^2\times B^2\) there is a chart \(\Gamma\) with \(\rho_S=\rho_\Gamma\), and conversely every chart \(\Gamma\) arises from a unique \(2\)-dimensional braid \(S\) up to isomorphism [1312.5388]. In the sphere-based language of branched coverings, an unoriented permutation-chart records a homomorphism into \(S_d\), whereas an oriented braid-chart records a homomorphism into \(B_d\); if a simple \(d\)-fold covering \(f:M^2\to S^2\) admits an embedded lift \(g:M^2\hookrightarrow D^2\times S^2\), one may choose \(\Gamma\) to be orientable so that its monodromy is the braid monodromy of \(g\) [1206.4744]. The monodromy description also admits a direct cut-and-paste realization. Starting with an \(S_d\)-chart in \(S^2\), one takes \(d\) copies of \(S^2\), cuts sheet \(j\) along those edges labeled \(i\) for which \(j=i\) or \(j=i+1\), and then glues the cut arcs on sheets \(i\) and \(i+1\) in reverse orientation. The result is a closed oriented surface \(M^2\) with a simple branched covering \(f:M^2\to S^2\), and in the braid-chart case the same construction can be viewed as an embedded surface in \(D^2\times S^2\) where each labeled edge introduces a half-twist in the \(D^2\)-fiber [1206.4744]. ## 3. Local moves, equivalence, and simplification Charts are subject to local modifications that preserve monodromy. In the formulation for \(2\)-dimensional braids, the basic moves include creation or annihilation of a \(2\)-gonal OVAL NEST, the CII-move for commuting distant labels, the CIII-move for the braid relation, and sliding black vertices past crossings or white vertices when labels permit [1312.5388]. These realize, respectively, local identities such as \[ \sigma_i\sigma_j=\sigma_j\sigma_i\qquad (|i-j|>1) \] and \[ \sigma_i\sigma_{i+1}\sigma_i\leftrightarrow \sigma_{i+1}\sigma_i\sigma_{i+1}. \] Two charts describe isomorphic or equivalent \(2\)-dimensional braids precisely when they are related by ambient isotopies together with a finite sequence of these chart moves [1312.5388]. For charts drawn on surface diagrams of surface-knots, the equivalence relation enlarges to include C-moves and Roseman moves. In this setting \((F,\Gamma)\) uniquely presents the \(2\)-dimensional braid up to ambient isotopy, C-moves, and Roseman moves [1503.00404]. The same paper develops an additional calculus based on adding \(1\)-handles equipped with chart loops. In the repeated-pattern case, a handle carrying \(m\) parallel copies of a braid \(b\) around the cocore and \(n\) copies around the core is denoted \[ h(m,n), \] and the basic \(1\)-handle moves include \[ h(m,n)\sim h(m,n\pm 2m), \] \[ 1(m,n)\sim 1(-n,m)\sim 1(n,-m), \] and the handle-slide relation \[ h(m,n)+h'(m',n')\sim hh'(m,n+n')+h'(m'-m,n'). \] These moves support Euclidean-algorithm-style simplifications of repeated-pattern braids and lead to normal forms involving a single nonzero \(m\)-parameter [1503.00404]. For general charts without branch points, the same handle technology becomes an unbraiding operation. By adding finitely many \(1\)-handles of the form \(1(\sigma_i,e)\) and \(1(e,e)\), one can eliminate all chart vertices and loops and reduce to \[ (F,\emptyset)+\sum_\lambda H_\lambda, \] where each \(H_\lambda\) carries only one chart-loop of type \(\sigma_i\) or is trivial. The weak unbraiding number \(u_w(F,\Gamma)\) and the unbraiding number \(u(F,\Gamma)\) quantify the minimal numbers of such handles, and a rough bound is \[ u_w(F,\Gamma)\le w(\Gamma)+2\,c(\Gamma)+N-1. \] If the chart has at least \(2(N-1)\) black vertices, one can finish the unbraiding with the standard handles \(\sum_{i=1}^{N-1}1(\sigma_i,e)\) plus trivial ones, leaving an unknotted chart consisting only of free edges [1503.00404]. In the degree-\(3\) chart of the \(2\)-twist-spun trefoil, a single added handle suffices: \[ u(S^2,\Gamma)=u_w(S^2,\Gamma)=1. \] ## 4. Charts for surface-links and chart-based invariants An \(n\)-chart \(\Gamma\subset D^2\) determines a simple surface braid \(S\subset D^1\times D^2\): over the complement of a neighborhood of \(\Gamma\), the projection is a trivial \(n\)-strand braid; black vertices become simple branch points; and a Hurwitz arc system from a base point to the black vertices records braid words that prescribe the local gluings. After identifying \(D^1\times S^2\subset \mathbb R^4\) with a tubular neighborhood of a standard \(2\)-sphere and capping off, one obtains a closed surface-link \[ F(\Gamma)\subset \mathbb R^4. \] Kamada’s theorem implies that every oriented surface-link arises, up to ambient isotopy, as the closure of some simple surface braid and hence from some \(n\)-chart [2304.05532]. Chart enumeration can therefore be used to organize families of surface-links. For \(4\)-charts with exactly two crossings and eight black vertices, \[ c(\Gamma)=2,\qquad b(\Gamma)=8, \] Nagase and Shima show that every c-minimal chart is, up to C-move equivalence and conjugation, one of the families \(\{T_k\}_{k\ge 0}\), \(\{T_k^*\}_{k\ge 0}\), or the exceptional chart \(T_0\); each has exactly two crossings, eight black vertices, and four Type-\(I_k\) IO-tangles of labels \(1\) or \(3\) [2304.05532]. | Chart family | Closure type | |---|---| | \(T_{2m}\), \(T_{2m}^*\) | Connected surface-links | | \(T_{2m-1}\), \(T_{2m-1}^*\), \(T_0\) | Exactly two connected components | To distinguish charts that are not separated by combinatorics alone, the paper uses quandle colorings. For a finite quandle \((Q,*)\), a coloring assigns quandle elements to the broken sheets of a generic projection, subject at each double point to \[ C(R_i)*C(R_j)=C(R_k), \] and the cardinality \(\#_Q(F)\) is an ambient isotopy invariant. For the dihedral quandle \(Q_N\), the exact counts include \[ \#_{Q_N}(F(T_0))=(N-1)^2+1, \] and \[ \#_{Q_N}(F(T_{2k}))=\#_{Q_N}(F(T_{2k}^*))=N. \] Moreover, for fixed \(k\), the two-component charts \(T_{2k-1}\) and \(T_{2\ell-1}\) with \(1\le \ell1\) leaves a Seifert surface for the branch link, and any simple \(2\)-fold branched covering \(f:M^3\to S^3\) admits an embedded lift \[ g:M^3\hookrightarrow D^2\times S^3\subset \mathbb R^5. \] For arbitrary \(d\), every simple \(d\)-fold branched covering admits at least an immersed lift \[ g:M^3\looparrowright D^2\times S^3, \] whose singularities are transverse double-point loops disjoint from the branch link [1206.4744]. The standard examples in the paper are the trefoil, whose \(2\)-fold branched cover is the lens space \(L(3,1)\), and the \(3\)-colored trefoil, whose \(3\)-fold cover is \(S^3\) again [1206.4744]. ## 6. Diagrammatic and algebraic descendants The chart formalism has acquired several later reinterpretations. In the theory of braid varieties, a positive braid word \[ \beta=\sigma_{i_1}\cdots \sigma_{i_\ell}\in Br_n^+ \] determines the affine variety \(V(\beta)=X_0(\beta;w_0)\), and planar graphs called weaves encode moves between braid words. A simplifying weave \(w\) from \(\gamma\) to the half-twist \(\Delta\) induces an injective map \[ M(w):\mathbb C^a\times (\mathbb C^*)^b\times X_0(\Delta;w_0)\cong \mathbb C^a\times (\mathbb C^*)^b\to X_0(\gamma;w_0), \] whose images give a cell decomposition of \(V(\gamma)\). For Demazure weaves one obtains maximal toric charts \[ T_w\cong (\mathbb C^*)^{\ell(\beta)}\hookrightarrow Y, \] and on each chart the holomorphic \(2\)-form has the log-canonical shape \[ \omega=\sum_{i