Weighted Cycles on Weaves Daping Weng Abstract . We introduce weighted cycles on weaves of general Dynkin types and define a skew-symmetrizable intersection pairing between weighted cycles. We prove that weighted cycles on a weave form a Laurent polynomial algebra and construct a quantization for this algebra using the skew-symmetric intersection pairing in the simply-laced case. We define merodromies along weighted cycles as functions on the decorated flag moduli space of the weave. We relate weighted cycles with cluster variables in a cluster algebra and prove that mutations of weighted cycles are compatible with mutations of cluster variables. 1. Introduction Weaves were introduced by Casals and Zaslow [ CZ22 ] as a graphical tool to describe a family of Legendrian surfaces living inside the 1 -jet space of a base surface. Soon after their introduction, weaves were adapted for the study cluster algebras, with many pieces of literature devoted to the development of weave-based descriptions of cluster structures, including [ Hug23 , ABL25 , CGGS24 , CLSBW23 ] . In particular, weaves were generalized to all Dynkin types in [ CGG+24 ] and played an important role in the construction of cluster structures on braid varieties. Under the generalization in loc. cit. , the original version of weaves by Casals and Zaslow, which can also be referred to as Legendrian weaves , would be categorized as Dynkin type A. In [ CW24 ] , Casals and the author gave a topological interpretation of cluster structures described by weaves of Dynkin type A, which can be summarized by the following table. Cluster Algebras Legendrian Weaves Cluster seed Exact Lagrangian surface S of a weave Quiver vertices \begin{array}[]{c}\text{Special collection of 1-cycles}\\ \text{on $S$ (also called Y-cycles)}\end{array} Quiver arrows \begin{array}[]{c}\text{Intersection pairing}\\ \text{between Y-cycles}\end{array} Cluster variables \begin{array}[]{c}\text{Merodromies along relative}\\ \text{1-cycles dual to Y-cycles}\end{array} Cluster mutation Polterovich surgery on S Table 1. Parallels between Legendrian weaves and their cluster algebras. Although Y-cycles together with their pairings and mutations have been generalized to weaves of general Dynkin types in [ CGG+24 ] , the rest of the parallels have been absent for general Dynkin types. In this article, we introduce weighted cycles on the base surface to fill this gap for general Dynkin types. To put it simply, we define a weighted chain to be a relative 1-chain on the base surface that intersects the weave generically, with the data of a weight (of the corresponding simply-connected Lie group) attached to each segment in the complement of the intersection with the weave; we also introduce homotopies of weighted chains and a condition on how to glue them together to form a weighted cycle. We define a commutative algebra \mathcal{W}(\mathfrak{w}) on the space of formal linear combinations of weighted cycles called the weighted cycle algebra , where the product is defined by stacking weighted cycles on top of each other diagrammatically. The weighted cycle algebra \mathcal{W}(\mathfrak{w}) is a topological incarnation of the Laurent polynomial ring of a cluster seed, as the following theorem indicates. Main Theorem 1 (Theorem 3.12 ) . The weighted cycle algebra \mathcal{W}(\mathfrak{w}) is a Laurent polynomial algebra of rank \tau+r(\beta-1) , where \tau is the number of trivalent vertices in \mathfrak{w} , r is the rank of the Dynkin type, and \beta is the number of boundary base points. Moreover, we define a homotopy-invariant skew-symmetrizable intersection pairing between weighted cycles, which allows us to introduce a quantization \mathbb{W}(\mathfrak{w}) of the weighted cycle algebra in the simply-laced case. We also describe weighted cycle representatives for Y-cycles, and prove the following. Main Theorem 2 (Theorem 3.15 ) . The intersection pairing between weighted cycle representatives of Y-cycles recovers the skew-symmetric pairing between Y-cycles. A weave also defines a flag moduli space on the base surface, for which we associate a flag (of the same Dynkin type) to each face of the weave and impose a relative position condition on each pair of flags on adjacent faces according to the color of the weave edge. In [ CW24 ] , Casals and the author defined a nowhere-vanishing function called merodromy for each relative cycle that intersects a Legendrian (Dynkin type A) weave generically. Simply speaking, the flag data give rise to a rank 1 local system on the weave surface; thus, given two non-zero vectors at the two ends of a relative 1-cycle, one can parallel transport the vector at the source to the other endpoint (target) of the relative cycle using the rank 1 local system, and then take the ratio with the vector sitting at the target; this ratio is the merodromy of the relative 1-cycle. Although a topological interpretation in terms rank 1 local systems on surfaces is no longer available for other Dynkin types, we manage to define a similar nowhere-vanishing function for each weighted cycle on the flag moduli space for weaves of a general Dynkin type. This result generalizes the aforementioned merodromy construction from [ CW24 ] . By choosing a suitable collection of weighted cycles, we can represent cluster variables as merodromies, as the following theorem implies. Main Theorem 3 (Theorem 3.35 ) . Merodromies of the positive weighted cycles transform according to the cluster mutation formula under a mutation of Y-cycle. Furthermore, we observe that weighted cycles share many similarities with webs [ Kup96 ] , which are generators of skein algebras: for example, weighted cycles and webs are both attaching weight data to a network of oriented curves, they both admit quantizations, and they both can give rise to elements in cluster algebras. Yet there is one key difference: the multiplication between weighted cycles is q -commutative, whereas the multiplication between webs obeys the skein relation. We believe that weighted cycles should be viewed as localizations of webs, and we further conjecture the following. In particular, the case G=\mathrm{SL}_{2} is in parallel with [ Mul16 ] . Conjecture 4 . Let \mathrm{Sk}(S,G) denote the quantum skein algebra associated with a simply-connected Lie group G on a surface S . Let \mathfrak{w} be a G -weave arising from an ideal triangulation on S . Then there exists a quantum algebra homomorphism \mathrm{Sk}(S,G)\rightarrow\mathbb{W}(\mathfrak{w}) , mapping each web W to a linear combination of weighted cycles that share the same topological support as W . \omega_{2} \omega_{2} \omega_{2} -\omega_{1} -\omega_{1}\quad \quad\omega_{1}-\omega_{2} \quad-\omega_{1} \omega_{2} \bullet Figure 1. Left: an \mathrm{SL}_{3} -web. Right: the corresponding weighted cycle. The article is structured as follows: Section 2 reviews some basic background on decorated flags and weaves; Section 3 gives the definitions of weighted cycles and related concepts, and proves our main theorems in the simply-laced case; Section 4 revisits some well-known constructions in cluster algebras and showcases how to describe them using weighted cycles; Section 5 goes over the construction of (co)weighted cycles and their skew-symmetrizable intersection pairing in the non-simply-laced case. 2. Preliminaries 2.1. Weaves and Y-cycles Weaves were first introduced in [ CZ22 ] and they were originally devised as combinatorial tools to describe Legendrian surfaces in 1-jet spaces. Weaves were later generalized to all Dynkin types in [ CGG+24 ] , and weaves of Dynkin type A were the ones associated with Legendrian surfaces. In this subsection, we briefly review some basics about weaves of simply-laced Dynkin types, and for simplicity, we mostly only consider weaves on a disk (with Subsection 4.6 as an exception). Recall that simply-laced Dynkin diagrams are classified into two infinite families (types A and D) with three exceptions ( \mathrm{E}_{6} , \mathrm{E}_{7} , and \mathrm{E}_{8} ). They are of the following forms. \bullet \bullet \bullet \bullet \cdots 1 2 n-1 n \mathrm{A}_{n} \bullet \bullet \bullet \cdots \bullet \bullet 1 2 n-2 n-1 n \mathrm{D}_{n} \bullet \bullet \bullet \bullet \bullet \bullet 1 2 3 4 5 6 \mathrm{E}_{6} \bullet \bullet \bullet \bullet \bullet \bullet \bullet 1 2 3 4 5 6 7 \mathrm{E}_{7} \bullet \bullet \bullet \bullet \bullet \bullet \bullet \bullet 1 2 3 4 5 6 7 8 \mathrm{E}_{8} Figure 2. Simply-laced Dynkin diagrams. Each Dynkin diagram defines a Coxeter group W , which is generated by simple reflections s_{i} that are in bijection with the vertices in the Dynkin diagram. These simple reflections satisfy the following relations: • s_{i}^{2}=e , • s_{i}s_{j}=s_{j}s_{i} if i\neq j are two non-adjacent vertices in the Dynkin diagram, • s_{i}s_{j}s_{i}=s_{j}s_{i}s_{j} if i\neq j are two adjacent vertices in the Dynkin diagram. The last two relations are also known as the braid relations . There is a faithful action of W on the Euclidean space \mathbb{R}^{n} by reflections over hyperplanes, with n being the number of vertices in the Dynkin diagram. To define such an action, we first fix a basis \{\alpha_{i}\} such that • (\alpha_{i},\alpha_{i})=2 , • (\alpha_{i},\alpha_{j})=0 if i\neq j are two non-adjacent vertices in the Dynkin diagram, • (\alpha_{i},\alpha_{j})=-1 if i\neq j are two adjacent vertices in the Dynkin diagram. Then the action of the simple reflection s_{i} is defined to be the reflection over the hyperplane orthogonal to \alpha_{i} ; in formula, the action of s_{i} is s_{i}.v:=v-(\alpha_{i},v)\alpha_{i}. These basis vectors \alpha_{i} ’s are called simple roots . Definition 2.1 . A weave of simply-laced Dynkin types (ADE) is a planar graph embedded in the disk with edges labeled by the simple reflections of that Dynkin type, such that each vertex is one of the three types listed in Figure 3 . A weave edge is said to be external if it is incident to the boundary of the disk; otherwise it is said to be internal . s_{i} s_{i} \bullet s_{i} s_{i} s_{i} s_{i} s_{j} s_{i} s_{j} if s_{i}s_{j}=s_{j}s_{i} s_{i} s_{j} s_{i} s_{j} s_{i} s_{j} if s_{i}s_{j}s_{i}=s_{j}s_{i}s_{j} Figure 3. Allowable vertices in a weave. Remark 2.2 . The bivalent weave vertex is a new vertex type that we are introducing in this article. It serves the purpose of resolving conflicting orientations on weave edges, which are needed to define merodromies later in Subsection 3.5 . Definition 2.3 . In addition to the definition of weaves above, we would like to introduce boundary base points : they are a finite collection of points along the boundary of the disk away from any external weave edges. We require each weave to have at least one boundary base point. The connected components of the complement of boundary base points are called boundary intervals . Definition 2.4 . Let \mathfrak{w} be a weave and let E(\mathfrak{w}) be the set of weave edges in \mathfrak{w} . A Y-cycle is a map \gamma:E(\mathfrak{w})\rightarrow\mathbb{Z}_{\geq 0} satisfying the following three conditions: • Between the two weave edges a and b incident to a bivalent weave vertex, \gamma(a)=\gamma(b) . • Among the three weave edges a,b,c incident to a trivalent weave vertex, the minimum of \gamma(a) , \gamma(b) , and \gamma(c) is achieved at least twice. • Among the four weave edges a,b,c,d incident to a tetravalent weave vertex (in a cyclic order), \gamma(a)=\gamma(c) and \gamma(b)=\gamma(d) . • Among the six weave edges a,b,c,d,e,f incident to a hexavalent weave vertex (in a cyclic order), \gamma(a)-\gamma(d)=\gamma(e)-\gamma(b)=\gamma(c)-\gamma(f) . The support of a Y-cycle is the subgraph of \mathfrak{w} spanned by the weave edges e with \gamma(e)>0 . The number \gamma(e) is called the mulitplicity of \gamma at e . A Y-cycle is unfrozen if \gamma(e)=0 for all external weave edges e ; a Y-cycle is frozen if it is not unfrozen. A set of Y-cycles is said to be linearly independent if they are linearly independent as functions on E(\mathfrak{w}) . Definition 2.5 . On a weave \mathfrak{w} , let Y(\mathfrak{w}) be the set of Y-cycles, V(\mathfrak{w}) be the set of vertices, and X(\mathfrak{w}) be the set of external weave edges. Let \frac{1}{2}\mathbb{Z}=\{0,\pm\frac{1}{2},\pm 1,\pm\frac{3}{2},\dots\} denote the set of half-integers. The intersection pairing between Y-cycles is a map \{\cdot,\cdot\}:Y(\mathfrak{w})\times Y(\mathfrak{w})\rightarrow\frac{1}{2}\mathbb{Z} defined by \{\gamma,\gamma^{\prime}\}:=\sum_{v\in V(\mathfrak{w})}\{\gamma,\gamma^{\prime}\}_{v}+\sum_{e,e^{\prime}\in X(\mathfrak{w})}\gamma(e)\gamma^{\prime}(e^{\prime})\{e,e^{\prime}\}. The terms in the first summation are defined by \{\gamma,\gamma^{\prime}\}_{v}=\left\{\begin{array}[]{ll}\det\begin{pmatrix}1&1&1\\ \gamma(a)&\gamma(b)&\gamma(c)\\ \gamma^{\prime}(a)&\gamma^{\prime}(b)&\gamma^{\prime}(c)\end{pmatrix}&\text{if $\hbox to49.68pt{\vbox to49.58pt{\pgfpicture\makeatletter\hbox{\hskip 24.84087pt\lower-28.65276pt\hbox to0.0pt{\lxSVG@begingroup@{_scopebegin=1} \lxSVG@begingroup@{stroke=#000000} \lxSVG@begingroup@{fill=#000000} \lxSVG@setlinewidth{\the\pgflinewidth}\lxSVG@begingroup@{stroke-width=0.4pt} \lx@inpgf@ignorespaces\nullfont\hbox to0.0pt{\lxSVG@begingroup@{_scopebegin=1} {}{{}}{} {}{}{}{}{{{}{}}}\lxSVG@begingroup@{_scopebegin=1} \color[rgb]{0,0,1}{}\lxSVG@stroke\lxSVG@drawpath@unclipped{M -34.1 19.69 L 0 0}{fill:none} \lx@inpgf@ignorespaces\hbox{\hbox{{\lxSVG@begingroup@{_scopebegin=1} {{}{}{{ {}{}}}{ {}{}} {{}{{\lx@inpgf@ignorespaces}}}{{}{\lx@inpgf@ignorespaces}}{}{{}{\lx@inpgf@ignorespaces}} {\lx@inpgf@ignorespaces }{{{{\lx@inpgf@ignorespaces}}\lxSVG@begingroup@{_scopebegin=1} \lxSVG@transformcm{1.0}{0.0}{0.0}{1.0}{-14.46626pt}{10.6462pt}\lxSVG@begingroup@{transform=matrix(1.0 0.0 0.0 1.0 -20.02 14.73)} \pgfsys@hbox{58}\lxSVG@closescope }}} \lxSVG@closescope }}} \lxSVG@closescope {}{{}}{} {}{}{}{}{{{}{}}}\lxSVG@begingroup@{_scopebegin=1} \color[rgb]{0,0,1}{}\lxSVG@stroke\lxSVG@drawpath@unclipped{M 34.1 19.69 L 0 0}{fill:none} \lx@inpgf@ignorespaces\hbox{\hbox{{\lxSVG@begingroup@{_scopebegin=1} {{}{}{{ {}{}}}{ {}{}} {{}{{\lx@inpgf@ignorespaces}}}{{}{\lx@inpgf@ignorespaces}}{}{{}{\lx@inpgf@ignorespaces}} {\lx@inpgf@ignorespaces }{{{{\lx@inpgf@ignorespaces}}\lxSVG@begingroup@{_scopebegin=1} \lxSVG@transformcm{1.0}{0.0}{0.0}{1.0}{9.67749pt}{10.6462pt}\lxSVG@begingroup@{transform=matrix(1.0 0.0 0.0 1.0 13.39 14.73)} \pgfsys@hbox{58}\lxSVG@closescope }}} \lxSVG@closescope }}} \lxSVG@closescope {}{{}}{} {}{}{}{}{{{}{}}}\lxSVG@begingroup@{_scopebegin=1} \color[rgb]{0,0,1}{}\lxSVG@stroke\lxSVG@drawpath@unclipped{M 0 0 L 0 -39.37}{fill:none} \lx@inpgf@ignorespaces\hbox{\hbox{{\lxSVG@begingroup@{_scopebegin=1} {{}{}{{{}{}}}{{}{}} {{}{{\lx@inpgf@ignorespaces}}}{{}{\lx@inpgf@ignorespaces}}{}{{}{\lx@inpgf@ignorespaces}} {\lx@inpgf@ignorespaces }{{{{\lx@inpgf@ignorespaces}}\lxSVG@begingroup@{_scopebegin=1} \lxSVG@transformcm{1.0}{0.0}{0.0}{1.0}{-7.86055pt}{-16.37915pt}\lxSVG@begingroup@{transform=matrix(1.0 0.0 0.0 1.0 -10.88 -22.66)} \pgfsys@hbox{58}\lxSVG@closescope }}} \lxSVG@closescope }}} \lxSVG@closescope {{}}\lx@inpgf@ignorespaces\hbox{\hbox{{\lxSVG@begingroup@{_scopebegin=1} {{}{}{{ }}{ } {{}{{\lx@inpgf@ignorespaces}}}{{}{\lx@inpgf@ignorespaces}}{}{{}{\lx@inpgf@ignorespaces}} {\lx@inpgf@ignorespaces }{{{{\lx@inpgf@ignorespaces}}\lxSVG@begingroup@{_scopebegin=1} \lxSVG@transformcm{1.0}{0.0}{0.0}{1.0}{3.533pt}{-7.83855pt}\lxSVG@begingroup@{transform=matrix(1.0 0.0 0.0 1.0 4.89 -10.85)} \pgfsys@hbox{58}\lxSVG@closescope }}} \lxSVG@closescope }}} \lxSVG@closescope {\lx@inpgf@ignorespaces}{\lx@inpgf@ignorespaces}{\lx@inpgf@ignorespaces}\hss}\lxSVG@discardpath\lxSVG@closescope \hss}}\lxSVG@closescope\endpgfpicture}}$}\\ &\\ \dfrac{1}{2}\left(\det\begin{pmatrix}1&1&1\\ \gamma(a)&\gamma(c)&\gamma(e)\\ \gamma^{\prime}(a)&\gamma^{\prime}(c)&\gamma^{\prime}(e)\end{pmatrix}+\det\begin{pmatrix}1&1&1\\ \gamma(b)&\gamma(d)&\gamma(f)\\ \gamma^{\prime}(b)&\gamma^{\prime}(d)&\gamma^{\prime}(f)\end{pmatrix}\right)&\text{if $\hbox to61.54pt{\vbox to69.2pt{\pgfpicture\makeatletter\hbox{\hskip 30.57631pt\lower-33.93854pt\hbox to0.0pt{\lxSVG@begingroup@{_scopebegin=1} \lxSVG@begingroup@{stroke=#000000} \lxSVG@begingroup@{fill=#000000} \lxSVG@setlinewidth{\the\pgflinewidth}\lxSVG@begingroup@{stroke-width=0.4pt} \lx@inpgf@ignorespaces\nullfont\hbox to0.0pt{\lxSVG@begingroup@{_scopebegin=1} {}{{}}{}{{}} {}{}\lxSVG@begingroup@{_scopebegin=1} \color[rgb]{0,0,1}{}\lxSVG@stroke\lxSVG@drawpath@unclipped{M -34.1 19.69 L 0 0}{fill:none} \lx@inpgf@ignorespaces\hbox{\hbox{{\lxSVG@begingroup@{_scopebegin=1} {{}{}{{ {}{}}}{ {}{}} {{}{{\lx@inpgf@ignorespaces}}}{{}{\lx@inpgf@ignorespaces}}{}{{}{\lx@inpgf@ignorespaces}} {\lx@inpgf@ignorespaces }{{{{\lx@inpgf@ignorespaces}}\lxSVG@begingroup@{_scopebegin=1} \lxSVG@transformcm{1.0}{0.0}{0.0}{1.0}{-26.80464pt}{17.75938pt}\lxSVG@begingroup@{transform=matrix(1.0 0.0 0.0 1.0 -37.09 24.57)} \pgfsys@hbox{58}\lxSVG@closescope }}} \lxSVG@closescope }}} \lxSVG@closescope {}{{}}{}{{}} {}{}\lxSVG@begingroup@{_scopebegin=1} \color[rgb]{0,0,1}{}\lxSVG@stroke\lxSVG@drawpath@unclipped{M 34.1 19.69 L 0 0}{fill:none} \lx@inpgf@ignorespaces\hbox{\hbox{{\lxSVG@begingroup@{_scopebegin=1} {{}{}{{ {}{}}}{ {}{}} {{}{{\lx@inpgf@ignorespaces}}}{{}{\lx@inpgf@ignorespaces}}{}{{}{\lx@inpgf@ignorespaces}} {\lx@inpgf@ignorespaces }{{{{\lx@inpgf@ignorespaces}}\lxSVG@begingroup@{_scopebegin=1} \lxSVG@transformcm{1.0}{0.0}{0.0}{1.0}{21.99792pt}{17.75938pt}\lxSVG@begingroup@{transform=matrix(1.0 0.0 0.0 1.0 30.44 24.57)} \pgfsys@hbox{58}\lxSVG@closescope }}} \lxSVG@closescope }}} \lxSVG@closescope {}{{}}{}{{}} {}{}\lxSVG@begingroup@{_scopebegin=1} \color[rgb]{1,0,0}{}\lxSVG@stroke\lxSVG@drawpath@unclipped{M 0 39.37 L 0 0}{fill:none} \lx@inpgf@ignorespaces\hbox{\hbox{{\lxSVG@begingroup@{_scopebegin=1} {{}{}{{ {}{}}}{ {}{}} {{}{{\lx@inpgf@ignorespaces}}}{{}{\lx@inpgf@ignorespaces}}{}{{}{\lx@inpgf@ignorespaces}} {\lx@inpgf@ignorespaces }{{{{\lx@inpgf@ignorespaces}}\lxSVG@begingroup@{_scopebegin=1} \lxSVG@transformcm{1.0}{0.0}{0.0}{1.0}{3.533pt}{24.98055pt}\lxSVG@begingroup@{transform=matrix(1.0 0.0 0.0 1.0 4.89 34.57)} \pgfsys@hbox{58}\lxSVG@closescope }}} \lxSVG@closescope }}} \lxSVG@closescope {}{{}}{} {}{}{{}}\lxSVG@begingroup@{_scopebegin=1} \color[rgb]{0,0,1}{}\lxSVG@stroke\lxSVG@drawpath@unclipped{M 0 0 L 0 -39.37}{fill:none} \lx@inpgf@ignorespaces\hbox{\hbox{{\lxSVG@begingroup@{_scopebegin=1} {{}{}{{ {}{}}}{ {}{}} {{}{{\lx@inpgf@ignorespaces}}}{{}{\lx@inpgf@ignorespaces}}{}{{}{\lx@inpgf@ignorespaces}} {\lx@inpgf@ignorespaces }{{{{\lx@inpgf@ignorespaces}}\lxSVG@begingroup@{_scopebegin=1} \lxSVG@transformcm{1.0}{0.0}{0.0}{1.0}{3.533pt}{-30.60553pt}\lxSVG@begingroup@{transform=matrix(1.0 0.0 0.0 1.0 4.89 -42.35)} \pgfsys@hbox{58}\lxSVG@closescope }}} \lxSVG@closescope }}} \lxSVG@closescope {}{{}}{} {}{}{{}}\lxSVG@begingroup@{_scopebegin=1} \color[rgb]{1,0,0}{}\lxSVG@stroke\lxSVG@drawpath@unclipped{M 0 0 L 34.1 -19.69}{fill:none} \lx@inpgf@ignorespaces\hbox{\hbox{{\lxSVG@begingroup@{_scopebegin=1} {{}{}{{ {}{}}}{ {}{}} {{}{{\lx@inpgf@ignorespaces}}}{{}{\lx@inpgf@ignorespaces}}{}{{}{\lx@inpgf@ignorespaces}} {\lx@inpgf@ignorespaces }{{{{\lx@inpgf@ignorespaces}}\lxSVG@begingroup@{_scopebegin=1} \lxSVG@transformcm{1.0}{0.0}{0.0}{1.0}{21.65474pt}{-24.70383pt}\lxSVG@begingroup@{transform=matrix(1.0 0.0 0.0 1.0 29.96 -34.18)} \pgfsys@hbox{58}\lxSVG@closescope }}} \lxSVG@closescope }}} \lxSVG@closescope {}{{}}{} {}{}{{}}\lxSVG@begingroup@{_scopebegin=1} \color[rgb]{1,0,0}{}\lxSVG@stroke\lxSVG@drawpath@unclipped{M 0 0 L -34.1 -19.69}{fill:none} \lx@inpgf@ignorespaces\hbox{\hbox{{\lxSVG@begingroup@{_scopebegin=1} {{}{}{{ {}{}}}{ {}{}} {{}{{\lx@inpgf@ignorespaces}}}{{}{\lx@inpgf@ignorespaces}}{}{{}{\lx@inpgf@ignorespaces}} {\lx@inpgf@ignorespaces }{{{{\lx@inpgf@ignorespaces}}\lxSVG@begingroup@{_scopebegin=1} \lxSVG@transformcm{1.0}{0.0}{0.0}{1.0}{-27.2433pt}{-24.70381pt}\lxSVG@begingroup@{transform=matrix(1.0 0.0 0.0 1.0 -37.7 -34.18)} \pgfsys@hbox{58}\lxSVG@closescope }}} \lxSVG@closescope }}} \lxSVG@closescope {{}}\lx@inpgf@ignorespaces\hbox{\hbox{{\lxSVG@begingroup@{_scopebegin=1} {{}{}{{{}{}}}{{}{}} {{}{{\lx@inpgf@ignorespaces}}}{{}{\lx@inpgf@ignorespaces}}{}{{}{\lx@inpgf@ignorespaces}} {\lx@inpgf@ignorespaces }{{{{\lx@inpgf@ignorespaces}}\lxSVG@begingroup@{_scopebegin=1} \lxSVG@transformcm{1.0}{0.0}{0.0}{1.0}{-8.739pt}{-2.15277pt}\lxSVG@begingroup@{transform=matrix(1.0 0.0 0.0 1.0 -12.09 -2.98)} \pgfsys@hbox{58}\lxSVG@closescope }}} \lxSVG@closescope }}} \lxSVG@closescope {\lx@inpgf@ignorespaces}{\lx@inpgf@ignorespaces}{\lx@inpgf@ignorespaces}\hss}\lxSVG@discardpath\lxSVG@closescope \hss}}\lxSVG@closescope\endpgfpicture}}$}\\ &\\ 0&\text{otherwise}.\end{array}\right. The bracket \{e,e^{\prime}\} in the second summation is skew-symmetric and is 0 unless both e and e^{\prime} are on the same boundary interval. When they are on the same boundary interval, let us assume without loss of generality that e^{\prime} preceeds e in the clockwise direction, separated by external weave edges of colors s_{i_{1}},s_{i_{2}},\dots,s_{i_{k-1}} in the clockwise direction (as in Figure 4 ). Suppose e is of color s_{i_{0}} and e^{\prime} is of color s_{i_{k}} . We then define \{e,e^{\prime}\}=\frac{1}{2}(\alpha_{i_{0}},s_{i_{1}}\cdots s_{i_{k-1}}.\alpha_{i_{k}}), where \alpha_{i_{0}} and \alpha_{i_{k}} are simple roots. e s_{i_{0}} e^{\prime} s_{i_{k}} s_{i_{1}} s_{i_{k-1}} \cdots Figure 4. Intersection pairing between frozen Y-cycles. Lemma 2.6 . At a hexavalent weave vertex, \displaystyle\det\begin{pmatrix}1&1&1\\ \gamma(a)&\gamma(c)&\gamma(e)\\ \gamma^{\prime}(a)&\gamma^{\prime}(c)&\gamma^{\prime}(e)\end{pmatrix}= \displaystyle\det\begin{pmatrix}1&1&1\\ \gamma(b)&\gamma(d)&\gamma(f)\\ \gamma^{\prime}(b)&\gamma^{\prime}(d)&\gamma^{\prime}(f)\end{pmatrix} \displaystyle= \displaystyle\frac{1}{2}\left(\det\begin{pmatrix}1&1&1\\ \gamma(c)&\gamma(b)&\gamma(a)\\ \gamma^{\prime}(c)&\gamma^{\prime}(b)&\gamma^{\prime}(a)\end{pmatrix}-\det\begin{pmatrix}1&1&1\\ \gamma(d)&\gamma(e)&\gamma(f)\\ \gamma^{\prime}(d)&\gamma^{\prime}(e)&\gamma^{\prime}(f)\end{pmatrix}\right). Thus, the intersection pairing between Y-cycles at a hexavalent weave vertex can be replaced by either of the determinants in this lemma, and either of them is equal to the pairing given in [ CGG+24 ] . Proof. On the one hand, let u=\gamma(a)-\gamma(d)=\gamma(e)-\gamma(b)=\gamma(c)-\gamma(f) and let u^{\prime} be defined similarly for \gamma^{\prime} . Then \det\begin{pmatrix}1&1&1\\ \gamma(b)&\gamma(d)&\gamma(f)\\ \gamma^{\prime}(b)&\gamma^{\prime}(d)&\gamma^{\prime}(f)\end{pmatrix}=\det\begin{pmatrix}1&1&1\\ \gamma(b)+u&\gamma(d)+u&\gamma(f)+u\\ \gamma^{\prime}(b)+u^{\prime}&\gamma^{\prime}(d)+u^{\prime}&\gamma^{\prime}(f)+u^{\prime}\end{pmatrix}=\det\begin{pmatrix}1&1&1\\ \gamma(a)&\gamma(c)&\gamma(e)\\ \gamma^{\prime}(a)&\gamma^{\prime}(c)&\gamma^{\prime}(e)\end{pmatrix}. On the other hand, note that \displaystyle\det\begin{pmatrix}1&1&1\\ \gamma(c)&\gamma(b)&\gamma(a)\\ \gamma^{\prime}(c)&\gamma^{\prime}(b)&\gamma^{\prime}(a)\end{pmatrix}-\det\begin{pmatrix}1&1&1\\ \gamma(d)&\gamma(e)&\gamma(f)\\ \gamma^{\prime}(d)&\gamma^{\prime}(e)&\gamma^{\prime}(f)\end{pmatrix} \displaystyle= \displaystyle\gamma(c^{\prime})(\gamma(a)-\gamma(b))+\gamma^{\prime}(b)(\gamma(c)-\gamma(a))+\gamma^{\prime}(a)(\gamma(b)-\gamma(c)) \displaystyle-\gamma^{\prime}(d)(\gamma(f)-\gamma(e))-\gamma^{\prime}(e)(\gamma(d)-\gamma(f))-\gamma^{\prime}(f)(\gamma(e)-\gamma(d)) \displaystyle= \displaystyle\gamma(c^{\prime})\gamma(a)+\gamma(b)(\gamma^{\prime}(a)-\gamma^{\prime}(c))+\gamma^{\prime}(b)(\gamma(c)-\gamma(a))-\gamma^{\prime}(a)\gamma(c) \displaystyle-\gamma^{\prime}(d)\gamma(f)+\gamma(e)(\gamma^{\prime}(d)-\gamma^{\prime}(f))-\gamma^{\prime}(e)(\gamma(d)-\gamma(f))+\gamma^{\prime}(f)\gamma(d) \displaystyle= \displaystyle\gamma(c^{\prime})\gamma(a)+\gamma(b)(\gamma^{\prime}(f)-\gamma^{\prime}(d))+\gamma^{\prime}(b)(\gamma(f)-\gamma(d))-\gamma^{\prime}(a)\gamma(c) \displaystyle-\gamma^{\prime}(d)\gamma(f)+\gamma(e)(\gamma^{\prime}(a)-\gamma^{\prime}(c))-\gamma^{\prime}(e)(\gamma(a)-\gamma(c))+\gamma^{\prime}(f)\gamma(d) \displaystyle= \displaystyle\det\begin{pmatrix}1&1&1\\ \gamma(a)&\gamma(c)&\gamma(e)\\ \gamma^{\prime}(a)&\gamma^{\prime}(c)&\gamma^{\prime}(e)\end{pmatrix}+\det\begin{pmatrix}1&1&1\\ \gamma(b)&\gamma(d)&\gamma(f)\\ \gamma^{\prime}(b)&\gamma^{\prime}(d)&\gamma^{\prime}(f)\end{pmatrix}.\qed Definition 2.7 . There are certain moves between weaves that are considered equivalences (Figure 5 ). Weaves that are related by weave equivalences are said to be equivalent . Note that there is a natural bijection between Y-cycles on equivalent weaves: the correspondence is dictated by the multiplicities on the surrounding weave edges together with the conditions on multiplicities at the weave vertices. \overset{\text{I}}{\longleftrightarrow} \overset{\text{II}}{\longleftrightarrow} \overset{\text{III}}{\longleftrightarrow} \overset{\text{IV}}{\longleftrightarrow} \overset{\text{V}}{\longleftrightarrow} \overset{\text{VI}}{\longleftrightarrow} \bullet Figure 5. Weave equivalences. Definition 2.8 . An unfrozen Y-cycle \gamma is called a short I-cycle if there is a unique internal weave edge e such that \gamma(e)\neq 0 . Given a weave \mathfrak{w} and a short I-cycle \gamma on \mathfrak{w} , we can perform a mutation at \gamma to produce a new weave \mathfrak{w}^{\prime} . In the new weave \mathfrak{w}^{\prime} , the short I-cycle \gamma is replaced by a new short I-cycle \gamma^{\prime} . Note that two weaves differ by a mutation are not equivalent to each other. \gamma \mathfrak{w} \longleftrightarrow \gamma^{\prime} \mathfrak{w}^{\prime} Figure 6. Mutation at a short I-cycle 2.2. Decorated Flags, Distances, and Flag Moduli Spaces Let us recall a few definitions and constructions for decorated flags in this subsection. Let G be a simply-connected simple linear algebraic group. For simplicity, we assume that G is of simply-laced Dynkin type (ADE) here; the non-simply-laced case will be covered in Section 5 . Let us fix a Borel subgroup B\subset G and let N:=[B,B] be its maximal unipotent subgroup. The flag variety \mathcal{B} can be identified with the quotient space G/B , whose elements are hence called flags or undecorated flags . The decorated flag variety is defined to be the quotient space \mathcal{A}:=G/N , whose elements are called decorated flags . There is a natural projection map \pi:\mathcal{A}\rightarrow\mathcal{B} . Picking a decorated flag that maps to a particular flag is called choosing a decoration . If we fix a maximal torus T\subset B , the space of decorations over a fixed flag is isomorphic to T , as a result of the Levi decomposition B=TN . From the fixed maximal torus T we get a Weyl group W:=N_{G}T/T , which happens to be the Coxeter group defined by the Dynkin type (this is how simple linear algebraic groups are classified). The Weyl group W acts on T and hence also on its weight lattice X^{*}(T):=\mathrm{Hom}(T,\mathbb{G}_{m}) . The induced action of W on the linearization X^{*}(T)\underset{\mathbb{Z}}{\otimes}\mathbb{R} coincides with the faithful action of W on the Euclidean space \mathbb{R}^{n} . In particular, the choice of Borel subgroup B\supset T gives rise to a basis \{\omega_{i}\} of X^{*}(T) called the fundamental weights , which is dual to the basis of simple roots \{\alpha_{i}\} in the sense that (\alpha_{i},\omega_{j})=\delta_{ij} . A reduced word for a Weyl group element w is a product with the fewest Coxeter generators that multiply to w . The number of Coxeter generators in a reduced word of w is called the length of w . Any two reduced words of the same Weyl group element w can be transformed into each other by using only the braid relations. By fixing a collection of Chevalley generators of G , we obtain for each simple root \alpha_{i} a group homomorphism \varphi_{i}:\mathrm{SL}_{2}\rightarrow G . For each Coxeter generator s_{i} of the Weyl group, we define two lifts \overline{s}_{i}=\varphi_{i}\begin{pmatrix}0&-1\\ 1&0\end{pmatrix}\quad\text{and}\quad\overline{\overline{s}}_{i}=\varphi_{i}\begin{pmatrix}0&1\\ -1&0\end{pmatrix}. Note that both lifts satisfy the braid relations, and therefore we can use any reduced word of w to define lifts \overline{w} and \overline{\overline{w}} for a Weyl group element w . The Bruhat decomposition of G states that G=\bigsqcup_{w\in W}BwB . It can be further refined to G=\bigsqcup_{w\in W}N\overline{w}TN=\bigsqcup_{w\in W}N\overline{\overline{w}}TN. Definition 2.9 . For a pair of flags xB and yB , the Tits distance w(xB,yB) is the unique Weyl group element w such that x^{-1}y\in BwB . Goncharov and Shen introduced an h distance for a pair decorated flags in [ GS15 ] . We will use two versions of their h distances in this article. Definition 2.10 . For a pair of decorated flags xN and yN , the h_{+} distance h_{+}(xN,yN) is defined to be the unique element h\in T such that x^{-1}y\in N\overline{w}hN , and the h_{-} distance h_{-}(xN,yN) is defined to be the unique element h\in T such that x^{-1}y\in N\overline{\overline{w}}hN . Lemma 2.11 . Tits distance between flags and both versions of h distances between decorated flags are invariant under the respective diagonal G -actions. Proof. It follows from the fact that these distances only depend on the product x^{-1}y , which is invariant under the diagonal G -actions g.(xB,yB):=(gxB,gyB) and g.(xN,yN):=(gxN,gyN) . ∎ Recall that the coroot lattice X_{*}(T):=\mathrm{Hom}(\mathbb{G}_{m},T) is the lattice dual to X^{*}(T) . The simple coroots \{\alpha_{i}^{\vee}\} is the basis of X_{*}(T) dual to \{\omega_{i}\} . We will use exponentiation notation p^{\alpha^{\vee}} to denote the image of p\in\mathbb{G}_{m} under the an element \alpha^{\vee} in the coroot lattice; in particular, for a coroot \alpha_{i}^{\vee} , p^{\alpha_{i}^{\vee}}=\varphi_{i}\begin{pmatrix}p&0\\ 0&p^{-1}\end{pmatrix} for the group homomorphism \varphi_{i}:\mathrm{SL}_{2}\rightarrow G . Lemma 2.12 . Suppose x^{-1}y\in Bs_{i}B . Then h_{\pm}(xN,yN)=(-1)^{\alpha_{i}^{\vee}}h_{\mp}(xN,yN) . Proof. It follows from the fact that \overline{s}_{i}=\overline{\overline{s}}_{i}\cdot(-1)^{\alpha_{i}^{\vee}} . ∎ Recall from [ CZ22 , CW24 , CGG+24 ] that for a weave \mathfrak{w} on a disk, the flag moduli space \mathcal{M}(\mathfrak{w}) is the moduli space of flag configurations on \mathfrak{w} , i.e., we associate a flag (i.e., an element of G/B ) for each face of \mathfrak{w} , such that if two faces are separated by a weave edge of color s_{i} , their associated flags are in relative position s_{i} ; then we quotient these flag data by the global action of G . Definition 2.13 . With a choice of orientation on each external weave edge of \mathfrak{w} , we can further define a framed flag moduli space \mathcal{M}_{\mathrm{fr}}(\mathfrak{w}) : in addition to associating a flag with each face of \mathfrak{w} , we also associate a decorated flag (i.e., an element in G/N ) with each intersection of a face F and a boundary interval C , such that the following conditions are satisfied: • if F\cap C is associated with the decorated flag xN , then the flag at F is xB ; • if F and F^{\prime} are two boundary faces separated by a weave edge of color s_{i} that is a source (resp. sink) (left (resp. right) picture in Figure 7 ), and xN and x^{\prime}N are the decorated flags associated with the boundaries of F and F^{\prime} , respectively, then x^{-1}x^{\prime}\in N\overline{s}_{i}N (resp. x^{-1}x^{\prime}\in N\overline{\overline{s}}_{i}N ). s_{i} x^{\prime}N xN x^{\prime}B xB s_{i} x^{\prime}N xN x^{\prime}B xB Figure 7. Decorated flags along the boundary of a weave. In particular, if a boundary face F contains a boundary base point, then the two decorated flags associated with the two adjacent boundary intervals must share the same underlying undecorated flag. Similar to \mathcal{M}(\mathfrak{w}) , we also need to quotient out by the global action of G . Note that there is a natural forgetful map \mathcal{M}_{\mathrm{fr}}(\mathfrak{w})\longrightarrow\mathcal{M}(\mathfrak{w}). Remark 2.14 . In general, \mathcal{M}(\mathfrak{w}) and \mathcal{M}_{\mathrm{fr}}(\mathfrak{w}) are Artin stacks. 3. Weighted Cycles and Their Merodromies 3.1. Definition of Weighted Cycles The chains and cycles in this article should be understood as 1 -chains and 1 -cycles; since we will not discuss any chains or cycles of other dimensions, for simplicity, we omit the prefix 1 . Definition 3.1 . A weighted chain on a weave \mathfrak{w} is an oriented path \eta on the plane that avoids all boundary base points, with no self-intersection, and intersecting \mathfrak{w} transversely, together with the assignment of a weight to each connected component in \eta\setminus\mathfrak{w} , such that the two weights adjacent to any edge colored by s_{i} are of the form \mu and s_{i}.\mu . We also adopt the convention that a weighted chain with a 0 weight is the same as removing that weighted chain. Definition 3.2 . We also consider weighted chains up to homotopies , which is a combination of local path homotopies and the following list of moves (see also Figure 8 ): (0) Pulling or pushing a U-turn through a weave edge. (1) Reversing the orientation and changing the weights to their opposite. (2) Trivial if it does not intersect any weave edges and homotopic to a small loop or a boundary interval. (3) Homotoping through a bivalent weave vertex up to a sign. (4) Homotoping through a trivalent weave vertex of color s_{i} if s_{i}.\mu=\mu . (5) Homotoping through a tetravalent weave vertex. (6) Homotoping through a hexavalent weave vertex. s_{i} \mu s_{i}.\mu \mu = s_{i} \mu s_{i} \mu s_{i}.\mu = s_{i} -\mu -s_{i}.\mu (0) (1) \mu = \mu =1 s_{i} s_{i} \bullet \mu\ =(-1)^{\left\langle\alpha_{i}^{\vee},\mu\right\rangle} s_{i}.\mu (2) (3) s_{i} s_{i} s_{i} \mu =1 if s_{i}.\mu=\mu s_{i} s_{i} s_{j} s_{j} \mu s_{i}.\mu s_{j}s_{i}.\mu s_{j}.\mu =1 if s_{i}s_{j}=s_{j}s_{i} s_{i} s_{j} s_{i} s_{j} s_{i} s_{j} \mu s_{i}.\mu s_{j}s_{i}.\mu s_{i}s_{j}s_{i}.\mu s_{j}.\mu s_{i}s_{j}.\mu =1 if s_{i}s_{j}s_{i}=s_{j}s_{i}s_{j} (4) (5) (6) Figure 8. Homotopies in the simply-laced case. Remark 3.3 . Note that in these homotopy moves, trivial homotopy is set to 1 instead of 0 ; this is because we are treating the weighted cycles multiplicatively instead of additively. This is particularly needed to make sense of the sign in homotopy move (3). Also, please note that there is no ambiguity in the sign factor in homotopy move (3) since \left\langle\alpha_{i}^{\vee},s_{i}.\mu\right\rangle=\left\langle s_{i}.\alpha_{i}^{\vee},\mu\right\rangle=-\left\langle\alpha_{i}^{\vee},\mu\right\rangle . Definition 3.4 . A weighted (relative) cycle on a weave \mathfrak{w} is a collection of non-intersecting weighted chains with endpoints that are one of the two cases below: • on the boundary of the disk away from the boundary base points; • a generic interior point of the disk such that, after possibly reversing orientations (homotopy move (1)) to make it a sink for all incident weighted chains, the sum of all nearby weights is 0 . We call this the balancing condition on the interior endpoint (Figure 9 ). A weighted cycle is called a weighted absolute cycle if the weighted cycle homotopy class has a representative with no endpoints on the boundary of the disk. \mu_{1} \mu_{2} \mu_{3} \mu_{4} \bullet \mu_{1}+\mu_{2}+\mu_{3}+\mu_{4}=0 Figure 9. An interior point where multiple weighted chains end. \bullet \bullet \omega_{1} -\omega_{1} 2\omega_{1} \omega_{1} -\omega_{1} \bullet \omega_{1} \omega_{1} -\omega_{1} -\omega_{1} Figure 10. Left: a weighted cycle on an \mathrm{SL}_{2} -weave. Right: a weighted absolute cycle on an \mathrm{SL}_{2} -weave. Recall that the boundary of the disk is decorated with base points, which cut the boundary of the disk into boundary intervals. In addition to path homotopies, we also allow (7) homotopies that move boundary endpoints of weighted cycles along boundary intervals. We also add the following additional homotopy moves for interior endpoints of weighted chains. (8) Adding/removing an interior endpoint. (9) Pushing an interior endpoint off the boundary of the disk. (10) Expand/contract adjacent interior endpoints. (11) Split/combine weighted chains between interior endpoints. (12) Moving an interior endpoint of weighted chains through a weave edge. s_{i} \mu = s_{i} s_{i}.\mu \mu \mu = \mu \mu \bullet (7) (8) \mu_{1} \mu_{k} \mu_{m} \mu_{k+1} \vdots \vdots \bullet = \mu_{1} \mu_{k} \vdots \mu_{1} \mu_{k} \mu_{m} \mu_{k+1} \vdots \vdots \bullet = \bullet \bullet \nu \mu_{1} \mu_{k} \mu_{m} \mu_{k+1} \vdots \vdots \nu=\mu_{1}+\cdots+\mu_{k}-\mu_{k+1}-\cdots-\mu_{m} (9) (10) \mu_{1}+\mu_{2} \bullet \bullet = \mu_{1} \mu_{2} \bullet \bullet \cdot \cdot \cdot s_{i}.\mu_{1} \mu_{1} \mu_{2} \mu_{k} s_{i} \bullet = \cdot \cdot \cdot s_{i}.\mu_{1} \mu_{2} \mu_{k} s_{i} s_{i}.\mu_{2} s_{i}.\mu_{k} \bullet (11) (12) Figure 11. Homotopies involving interior endpoints. Remark 3.5 . There is a special case of homotopy move (10), where the intermediate weight \nu=0 ; in this case, we delete the weighted chain in the middle and the move will effectively split the interior endpoint in two. 3.2. Intersection Pairing between Weighted Cycles In this subsection we define an intersection number between weighted cycles. Definition 3.6 . When two weighted chains \eta_{1} and \eta_{2} intersect at p , we define the sign of the intersection p , \mathrm{sign}_{p}(\eta_{1},\eta_{2}) , as in Figure 12 . If \eta_{1} and \eta_{2} do not intersect, we define \mathrm{sign}_{p}(\eta_{1},\eta_{2})=0 . \eta_{2} \eta_{1} \mathrm{sign}_{p}(\eta_{1},\eta_{2})=1 \eta_{2} \eta_{1} \mathrm{sign}_{p}(\eta_{1},\eta_{2})=-1 Figure 12. Signs of the intersection between two weighted chains. Moreover, if \eta_{1} and \eta_{2} have weights \mu_{1} and \mu_{2} at p , we define the local intersection number between \eta_{1} and \eta_{2} at p to be \{\eta_{1},\eta_{2}\}_{p}:=\mathrm{sign}_{p}(\eta_{1},\eta_{2})\cdot(\mu_{1},\mu_{2}), where (\mu_{1},\mu_{2}) is the inner product between weights. Definition 3.7 . For a boundary endpoint p of a weighted chain, we define \mathrm{sign}(p)=-1 if it is a source and \mathrm{sign}(p)=1 if it is a sink. For a boundary interval C and two boundary endpoints p_{1} and p_{2} of weighted chains, we define \mathrm{sign}_{C}(p_{1},p_{2})=1 if p_{1} precedes p_{2} in the clockwise direction, \mathrm{sign}_{C}(p_{1},p_{2})=-1 if p_{1} precedes p_{2} in the counterclockwise direction, and \mathrm{sign}_{C}(p_{1},p_{2})=0 if at least one of p_{1} and p_{2} is not on C . Lastly, suppose \mu_{1} and \mu_{2} are weights at the boundary endpoints p_{1} and p_{2} on the same boundary interval C , separated by external weave edges of colors s_{i_{1}},s_{i_{2}},\dots,s_{i_{k}} in the clockwise direction (as in Figure 13 ); we define the inner product between p_{1} and p_{2} to be I_{C}(\mu_{1},\mu_{2}):=\left\{\begin{array}[]{ll}(\mu_{1},s_{i_{1}}\cdots s_{i_{k}}.\mu_{2})&\text{if $\mu_{2}$ precedes $\mu_{1}$ in the clockwise direction,}\\ (\mu_{2},s_{i_{1}}\cdots s_{i_{k}}.\mu_{1})&\text{if $\mu_{1}$ precedes $\mu_{2}$ in the clockwise direction.}\end{array}\right. C p_{1} \mu_{1} p_{2} \mu_{2} s_{i_{1}} s_{i_{k}} \cdots Figure 13. Inner product between boundary endpoints. We define the local intersection number between weighted chains \eta_{1} and \eta_{2} at a boundary interval C to be \{\eta_{1},\eta_{2}\}_{C}:=\frac{1}{2}\sum_{p_{1}\in\partial\eta_{1}}\sum_{p_{2}\in\partial\eta_{2}}\mathrm{sign}(p_{1})\mathrm{sign}(p_{2})\mathrm{sign}_{C}(p_{1},p_{2})I_{C}(\mu_{1},\mu_{2}). Definition 3.8 . We define the (total) intersection number between two weighted chains \eta_{1} and \eta_{2} to be \{\eta_{1},\eta_{2}\}:=\sum_{p\in\eta_{1}\cap\eta_{2}}\{\eta_{1},\eta_{2}\}_{p}+\sum_{C}\{\eta_{1},\eta_{2}\}_{C}. It is not hard to see that \{\cdot,\cdot\} is skew-symmetric. We extend this intersection number linearly to an intersection number between weighted cycles. Proposition 3.9 . The intersection number between weighted cycles is invariant under homotopies. Proof. Since the intersection number is defined locally and the intersections are assumed to be in generic position, we can always isolate the intersections between weighted cycles away from most of the homotopy moves on the list. Thus, it suffices to only consider the following. (i) Path homotopies that create/cancel pairs of intersections between weighted cycles. (ii) Homotopy move (7) that moves boundary endpoints of weighted chains past each other. (iii) Homotopy move (7) that moves a boundary endpoint of a weighted cycle through an external weave edge. (iv) Homotopy move (1), which reverses the orientation of a weighted cycle and changes the weights to their opposite. (v) Homotopy move (9), which pushes an interior endpoint off the boundary of the disk. (vi) Pushing/pulling an intersection between weighted cycles through a weave edge. (vii) Pushing/pulling intersections between weighted cycles through an interior endpoint. \mu_{1} \mu_{2} \longleftrightarrow \mu_{1} \mu_{2} \mu_{1} p_{1} \mu_{2} p_{2} \longleftrightarrow \mu_{1} p_{1} p \mu_{2} p_{2} p_{1} \mu_{1} s_{i_{1}} s_{i_{k}} p_{2} \mu_{2} \cdots \longleftrightarrow p_{1} \mu_{1} s_{i_{1}} s_{i_{k}} p_{2} \mu_{2} \cdots (i) (ii) (iii) \mu_{1} \mu_{2} \longleftrightarrow -\mu_{1} \mu_{2} \bullet \cdots \cdots p_{1} \cdots p_{k} \mu_{1} \mu_{k} \mu_{k+1} \mu_{m} q \nu \longleftrightarrow \cdots p_{k+1} \cdots p_{m} \mu_{k+1} \mu_{m} q \nu (iv) (v) \mu_{1} s_{i}.\mu_{1} \mu_{2} s_{i}.\mu_{2} s_{i} \longleftrightarrow \mu_{1} s_{i}.\mu_{1} \mu_{2} s_{i}.\mu_{2} s_{i} \mu_{1} \mu_{k} \mu_{m} \mu_{k+1} \nu \vdots \vdots \bullet \longleftrightarrow \mu_{1} \mu_{k} \mu_{m} \mu_{k+1} \nu \vdots \vdots \bullet (vi) (vii) Figure 14. Invariance of intersection numbers under homotopies of weighted cycles. For (i), regardless of the orientations on the weighted chains, the new pair of intersection points have opposite signs. Thus, their combined contribution is (\mu_{1},\mu_{2})-(\mu_{1},\mu_{2})=0 . For (ii), the LHS has a local contribution \{p_{1},p_{2}\}_{C}=-\frac{1}{2}(\mu_{1},\mu_{2}) , and the RHS has a local contribution \{p_{1},p_{2}\}_{C}+\{\eta_{1},\eta_{2}\}_{p}=\frac{1}{2}(\mu_{1},\mu_{2})-(\mu_{1},\mu_{2}) , which is equal to the LHS. For (iii), the inner product I_{C}(p_{1},p_{2}) changes from (\mu_{1},s_{i_{1}}\cdots s_{i_{k}}.\mu_{2}) to (s_{i_{1}}.\mu_{1},s_{i_{2}}\cdots s_{i_{k}}.\mu_{2}) , which is equal to (\mu_{1},s_{i_{1}}\cdots s_{i_{k}}.\mu_{2}) by the invariance of inner product between weights under the Weyl group action. For (iv), the sign of the intersection flips, which cancels with the new minus sign on the weight, so the local contribution does not change. For (v), the combined contribution \sum_{i=1}^{k}\{q,p_{i}\} is equal to \sum_{i=1}^{k}(\mu_{i},\nu)=\left(\sum_{i=1}^{k}\mu_{i},\nu\right) , and the combined contribution \sum_{i=k+1}^{m}\{q,p_{i}\} is equal to -\sum_{i=k+1}^{m}(\mu_{i},\nu)=\left(-\sum_{i=k+1}^{m}\mu_{i},\nu\right)=\left(\sum_{i=1}^{k}\mu_{i},\nu\right) , so the combined contribution does not change. For (vi), regardless of the orientations on the weighted chains, the sign of the intersection does not change, and the pairing between weights becomes (s_{i}.\mu_{1},s_{i}.\mu_{2}) , which is equal to (\mu_{1},\mu_{2}) since the inner product between weights is invariant under the Weyl group action. For (vii), since \mu_{k+1}+\cdots+\mu_{m}=-\mu_{1}-\cdots-\mu_{k} , we have (\mu_{1},\nu)+\cdots+(\mu_{k},\nu)=-(\mu_{k+1},\nu)-\cdots-(\mu_{m},\nu) ; but then the signs of intersections between \mu_{k+1},\dots,\mu_{m} and \nu are opposite to those between \mu_{1},\dots,\mu_{k} and \nu . Thus the local contribution does not change either. ∎ 3.3. Weighted Cycle Algebra and Quantization Definition 3.10 . Given a weave \mathfrak{w} , we define the weighted (relative) cycle algebra \mathcal{W}(\mathfrak{w}) to be the commutative algebra over \mathbb{C} whose elements are formal linear combinations of homotopy classes of weighted relative cycles on \mathfrak{w} , with a multiplication defined by first superimposing weighted cycles on top of each other generically and then replacing each crossing with an interior endpoint of weighted chains. \mu_{1} \mu_{2} = \mu_{1} \mu_{2} \bullet Figure 15. Multiplication in weighted cycle algebras. Note that the empty weighted chain (e.g., the 0 ’s in Figure 8 ) is the multiplicative identity in \mathcal{W}(\mathfrak{w}) . Definition 3.11 . Recall that a weighted cycle is absolute if there is a representative whose weighted chains do not end at the boundary of \mathfrak{w} . Since the product of two weighted absolute chains is still absolute, weighted absolute chains form a subalgebra inside \mathcal{W}(\mathfrak{w}) , which we call the weighted absolute cycle algebra and denote by \mathcal{W}_{\mathrm{abs}}(\mathfrak{w}) Theorem 3.12 . Suppose \mathfrak{w} is a weave for a simply connected semisimple Lie group G of rank r . Let \beta be the number of boundary base points on the disk and let \tau be the number of trivalent weave vertices in \mathfrak{w} . The weighted cycle algebra \mathcal{W}(\mathfrak{w}) is a Laurent polynomial ring whose spectrum is an algebraic torus of dimension \tau+r(\beta-1) . Proof. It suffices to show that the weighted cycles form a \mathbb{Z} -lattice of rank \tau+r(\beta-1) . First, we claim that each time we add a boundary base point, we add a summand of \mathbb{Z}^{r} . To see this, let \eta be a small oriented path surrounding only the new base point. For each fundamental weight \omega_{i} with 1\leq i\leq r we define a weighted cycle \eta_{i} to be \eta together with the weight \omega_{i} . These are the basis for the new \mathbb{Z}^{r} summand. The statement is now reduced to showing that if there is only one boundary base point, the rank of the lattice of weighted cycles is \tau . We observe that we can remove all interior endpoints for weighted chains inside a weighted cycle by using homotopy moves. To see this, note that we can use homotopy move (1) and (8) to remove all bivalent interior endpoints of weighted chains, and then use homotopy move (10) to make all remaining interior endpoints trivalent. We then do an induction on the number of interior endpoints. There is nothing to show for the base case with no interior endpoints. Inductively, pick a trivalent interior endpoint and draw a curve C that goes from this interior endpoint to the boundary of the disk, intersecting only the interiors of the weighted chains inside the weighted cycle and the interiors of weave edges. By using homotopy move (8), we put a new bivalent interior endpoint at each intersection between C and the weighted cycle. By using homotopy moves (9) and (12), we can push these bivalent interior endpoints off the boundary of the disk one-by-one, which creates a path to push the chosen trivalent interior endpoint off the boundary of the disk as well. \bullet C \longrightarrow \bullet Figure 16. Removing a trivalent interior endpoint. With the assumption that there are no interior endpoints for weighted chains, each weighted chain in a weighted cycle must be an arc that goes from some boundary interval to another boundary interval (may possibly be the same one), or forms a closed circle. By adding a generic bivalent interior endpoint and using a move similar to Figure 16 , we can open up any closed circle into an arc as well. Thus, without loss of generality, we may assume that all weighted chains are pairwise non-intersecting arcs. Let us do an induction on the number of trivalent weave vertices \tau . For the base case where \tau=0 , all weighted chains can be homotoped to the unique boundary interval using homotopy moves (3), (5), and (6), and they can be further shown to be trivial by moving the boundary endpoints together via homotopy move (7) and then applying the homotopy move (2). Now inductively, since all weighted chains inside the weighted cycle are pairwise non-intersecting arcs, every connected component in the complement of the weighted cycle must contain some boundary interval of the disk. Let us fix one trivalent weave vertex v of color s_{i} inside one of these connected components. We choose a simple curve \gamma inside this connected component, going from one point on the boundary to another point on the boundary, intersecting the weave at generic positions and isolating the weave vertex v from the rest of the weave (see Figure 17 ). We may choose \gamma so that it also intersects a weave edge incident to v at a point p . Note that \gamma cuts the disk into two disks, with all weighted chains contained inside the disk without the trivalent weave vertex v . On the one hand, the smaller disk without the weave vertex v has one fewer trivalent weave vertex, and hence by induction, the lattice of weighted cycles on this smaller disk is of rank \tau-1 . On the other hand, by using homotopy move (4), most weighted chain with a support that is homotopic to \gamma can be further homotoped to the boundary of the original disk through the other smaller disk containing v . For an obstructed weighted chain, let \omega be its weight near the point p . Since \{\omega_{j}\} is a basis of the weight lattice, we can write \omega=\sum_{j}c_{j}\omega_{j} . By using homotopy move (11), we can split this weighted chain into multiple weighted chains with the same support and with weights c_{j}\omega_{j} near p . Because s_{i}.\omega_{j}=\omega_{j} for all j\neq i , all these newly split weighted chains can be homotoped through the smaller disk except the one with the weight c_{i}\omega_{i} and c_{i}\neq 0 . This adds 1 to the rank of the lattice of weighted cycles and hence the total rank is \tau . The induction is now complete. ∎ \gamma v p Figure 17. Isolating a single trivalent weave vertex. Since the weighted cycles are equipped with an intersection pairing, we can further quantize the weighted cycle algebra \mathcal{W}(\mathfrak{w}) by imposing the following relation on weighted cycles: \eta_{1}\eta_{2}=q^{\{\eta_{1},\eta_{2}\}}(\eta_{1}\#\eta_{2}), where (\eta_{1}\#\eta_{2}) denotes the weighted cycle that is the classical product between \eta_{1} and \eta_{2} . We denote the quantum cycle algebra by \mathbb{W}(\mathfrak{w}) . It is not hard to see that we can recover \mathcal{W}(\mathfrak{w}) from \mathbb{W}(\mathfrak{w}) by setting q=1 . 3.4. Weighted Cycle Representatives for Y-Cycles We can construct weighted cycle representatives of Y-cycles as follows. • Let \gamma be a Y-cycle. For each weave edge e in \gamma , we draw a relative chain on each side of e and orient them such that e is on the right side of the relative chain. • If e has color s_{i} , we assign the weight \gamma(e)\omega_{i} to each of the two relative chains. • Near a bivalent weave vertex, we connect the nearby weighted chains as in the top left picture of Figure 18 . • Near a trivalent weave vertex, we connect the nearby weighted chains as in the top right picture of Figure 18 , where \mu_{j}:=|a_{j-1}-a_{j+1}|\omega_{i}-[a_{j-1}-a_{j+1}]_{+}\alpha_{i} with the indices taken modulo 3 . (Recall that [n]_{+}:=\max\{n,0\} , and \alpha_{i} is the i th simple root.) • Near a tetravalent weave vertex, we connect the nearby weighted chains as in the bottom left picture of Figure 18 . • Near a hexavalent weave vertex, we connect the nearby weighted chains as in the bottom right picture of Figure 18 . The weight \mu_{i} is determined by the other two weights via the balancing condition. Note that the interior endpoint in the center should be perturbed slightly to meet the generic position requirement. s_{i} s_{i} \bullet a\omega_{i} a\omega_{i} \bullet a\omega_{i} a\omega_{i} \bullet \bullet a_{1}\omega_{i} a_{1}\omega_{i} \mu_{1} s_{i}.\mu_{1} s_{i} \bullet a_{2}\omega_{i} a_{2}\omega_{i} \mu_{2} s_{i}.\mu_{2} s_{i} \bullet a_{3}\omega_{i} a_{3}\omega_{i} \mu_{3} s_{i}.\mu_{3} s_{i} s_{i} s_{j} b\omega_{j} b\omega_{j} b\omega_{j} b\omega_{j} a\omega_{i} a\omega_{i} a\omega_{i} a\omega_{i} \bullet \bullet \bullet \bullet b\omega_{j} a\omega_{i} b\omega_{j} a\omega_{i} \bullet \bullet \bullet \bullet \bullet \bullet \mu_{1} \mu_{2} \mu_{3} \mu_{4} \mu_{5} \mu_{6} \bullet a_{1}\omega_{i} a_{1}\omega_{i} s_{i} a_{3}\omega_{i} a_{3}\omega_{i} s_{i} a_{5}\omega_{i} a_{5}\omega_{i} s_{i} a_{2}\omega_{j} a_{2}\omega_{j} s_{j} a_{4}\omega_{j} a_{4}\omega_{j} s_{j} a_{6}\omega_{j} a_{6}\omega_{j} s_{j} Figure 18. Weighted cycle representatives for Y-cycles. Proposition 3.13 . The balancing condition is satisfied at all interior endpoints of weighted chains inside a weighted cycle representative of a Y-cycle. Proof. The balancing condition on the two interior endpoints near a bivalent weave vertex is obvious. Next, let us consider an interior endpoint near a trivalent weave vertex, say the left one in the left picture of Figure 18 . The total weights incident to that endpoint is \displaystyle a_{2}\omega_{i}+s_{i}.\mu_{2}-a_{1}\omega_{i}-\mu_{1} \displaystyle= \displaystyle a_{2}\omega_{i}+|a_{1}-a_{3}|\omega_{i}-|a_{1}-a_{3}|\alpha_{i}+[a_{1}-a_{3}]_{+}\alpha_{i}-a_{1}\omega_{i}-|a_{3}-a_{2}|\omega_{i}+[a_{3}-a_{2}]_{+}\alpha_{i} \displaystyle= \displaystyle(a_{2}+|a_{1}-a_{3}|-a_{1}-|a_{3}-a_{2}|)\omega_{i}+(-|a_{1}-a_{3}|+[a_{1}-a_{3}]_{+}+[a_{3}-a_{2}]_{+})\alpha_{i}. Depending on the relations among the multiplicities a_{1},a_{2} , and a_{3} , we have three cases to consider. (i) a_{1}=a_{2}\leq a_{3} : in this case, the total incident weight is (a_{2}+(a_{3}-a_{1})-a_{1}-(a_{3}-a_{2}))\omega_{i}+(-(a_{3}-a_{1})+(a_{3}-a_{2}))\alpha_{i}=0. (ii) a_{1}=a_{3}\leq a_{2} : in this case, the total incident weight is (a_{2}-a_{1}-(a_{2}-a_{3}))\omega_{i}=0. (iii) a_{2}=a_{3}\leq a_{1} : in this case, the total incident weight is (a_{2}+(a_{1}-a_{3})-a_{1})\omega_{i}+(-(a_{1}-a_{3})+(a_{1}-a_{3}))\alpha_{i}=0. The balancing condition on the other two endpoints can be proved by symmetric arguments. The balancing condition on the four interior endpoints near a tetravalent weave vertex is obvious since there is an exact match of incoming and outgoing weighted chains at each interior endpoint. Lastly, the only non-obvious balancing condition near a hexavalent weave vertex is on the center interior endpoint. By perturbing it upward slightly, we get the following picture. \mu_{1} \mu_{2} \mu_{3} \mu_{4} \mu_{5} \mu_{6} \bullet Figure 19. Perturbing the center interior endpoint at a hexavalent vertex. Without loss of generality, let us assume that the blue weave edges are colored by s_{1} and the red weave edges are colored by s_{2} ; by construction, \mu_{i}=a_{i+1}\omega_{2}-a_{i}\omega_{1} for i=1,3,5 , and \mu_{i}=a_{i+1}\omega_{1}-a_{i}\omega_{2} for i=2,4,6 (indices modulo 6 ). Then the sum of the weights incident to the center interior endpoint is \displaystyle s_{1}.\mu_{1}+s_{1}s_{2}.\mu_{2}+s_{2}s_{1}s_{2}.\mu_{3}+s_{2}s_{1}.\mu_{4}+s_{2}.\mu_{5}+\mu_{6} \displaystyle= \displaystyle a_{2}\omega_{2}-a_{1}(\omega_{1}-\alpha_{1})+a_{3}(\omega_{1}-\alpha_{1})-a_{2}(\omega_{2}-\alpha_{1}-\alpha_{2})+a_{4}(\omega_{2}-\alpha_{1}-\alpha_{2})-a_{3}(\omega_{1}-\alpha_{1}-\alpha_{2}) \displaystyle a_{5}(\omega_{1}-\alpha_{1}-\alpha_{2})-a_{4}(\omega_{2}-\alpha_{2})-+a_{6}(\omega_{2}-\alpha_{2})-a_{5}\omega_{1}+a_{1}\omega_{1}-a_{6}\omega_{2}- \displaystyle= \displaystyle(a_{1}+a_{2}-a_{4}-a_{5})\alpha_{1}+(a_{2}+a_{3}-a_{5}-a_{6})\alpha_{2}. Both coefficients vanish due to the condition that a_{1}-a_{4}=a_{3}-a_{6}=a_{5}-a_{2} . ∎ Furthermore, we can recover the intersection pairing between Y-cycles from the intersection pairing between their weighted cycle representatives. The following lemma is useful when computing the intersection pairing between complicated weighted cycles. Lemma 3.14 . Let C be a simple closed curve on the disk intersecting the weave \mathfrak{w} at generic positions. Suppose \mathfrak{w} does not have any trivalent vertices in the region enclosed by C . Let \eta_{1} and \eta_{2} be two weighted cycles and let p be a point on C that is away from \eta_{1} , \eta_{2} , and any weave edges in \mathfrak{w} . Then the contribution to \{\eta_{1},\eta_{2}\} from the region enclosed by C is equal to \{\eta^{\prime}_{1},\eta^{\prime}_{2}\}_{C\setminus\{p\}} , where \eta^{\prime}_{i} is the truncation of \eta_{i} in an outward tubular neighborhood of C . Proof. Note that the region enclosed by C is also a disk. Since the restriction of \mathfrak{w} to this smaller disk contains no trivalent weave vertices, and there is only one boundary base point p , by Theorem 3.12 , all weighted cycles inside this smaller disk are trivial. Let \eta^{\prime\prime}_{i} be the truncation of \eta_{i} inside this smaller disk. It follows from the discussion above that \{\eta^{\prime\prime}_{1},\eta^{\prime\prime}_{2}\}=0 . But the intersection pairing \{\eta^{\prime\prime}_{1},\eta^{\prime\prime}_{2}\} is the sum of the local contribution to the original intersection pairing \{\eta_{1},\eta_{2}\} together with the intersection pairing between \eta^{\prime\prime}_{1} and \eta^{\prime\prime}_{2} along the boundary interval C\setminus\{p\} . Thus, we can conclude that the local contribution to the original intersction pairing \{\eta_{1},\eta_{2}\} is -\{\eta^{\prime\prime}_{1},\eta^{\prime\prime}_{2}\}_{C\setminus\{p\}}=\{\eta^{\prime}_{1},\eta^{\prime}_{2}\}_{C\setminus\{p\}} . ∎ \eta_{1} \eta_{2} \mu_{2} \mu_{1} \bullet p C C\setminus\{p\} \mu_{1} \mu_{2} \mu_{1} \mu_{2} Figure 20. Demonstration of Lemma 3.14 : the local contribution in the left picture is (\mu_{1},\mu_{2}) , which is equal to the boundary interval contribution in the right picture. Theorem 3.15 . Let \eta_{1} and \eta_{2} be the weighted cycle representatives of Y-cycles \gamma_{1} and \gamma_{2} , respectively. Then the intersection pairing \{\gamma_{1},\gamma_{2}\} is equal to the intersection pairing \{\eta_{1},\eta_{2}\} . Proof. We can draw the weighted cycle representative \eta_{2} in a tubular neighborhood closer to the weave edges that \eta_{1} so that \eta_{1} and \eta_{2} only intersect near the weave vertices. Then it suffices to check the intersection pairings between \eta_{1} and \eta_{2} at each weave vertex and on each boundary interval and then compare the results with the intersection pairing contribution to \{\gamma_{1},\gamma_{2}\} from weave vertices and external edges. There are no intersections between \eta_{1} and \eta_{2} occurring near a bivalent weave vertex. At a trivalent weave vertex, by construction, all intersections must occur between the \eta_{2} weighted chains parallel to weave edges and the \eta_{1} weighted chains perpendicular to weave edges. Let a_{1},a_{2},a_{3} be the multiplicities for \gamma_{1} and let a^{\prime}_{1},a^{\prime}_{2},a^{\prime}_{3} be the multiplicities for \gamma_{2} . Then at each weave edge, we will have an intersection pairing contribution of the form (a^{\prime}_{j}\omega_{i},s_{i}.\mu_{j})-(a^{\prime}_{j}\omega_{i},\mu_{j})=(a^{\prime}_{j}\omega_{i},s_{i}.\mu_{j}-\mu_{j})=a^{\prime}_{j}\left(2[a_{j-1}-a_{j+1}]_{+}-|a_{j-1}-a_{j+1}|\right)=a^{\prime}_{j}(a_{j-1}-a_{j+1}). Thus, the total contribution locally near the trivalent weave vertex is \sum_{j=1}^{3}a^{\prime}_{j}(a_{j-1}-a_{j+1})=\det\begin{pmatrix}1&1&1\\ a_{1}&a_{2}&a_{3}\\ a^{\prime}_{1}&a^{\prime}_{2}&a^{\prime}_{3}\end{pmatrix}. At a tetravalent weave vertex, it is not hard to see that all intersections between \eta_{1} and \eta_{2} come in canceling pairs and therefore the total contribution locally near a tetravalent weave vertex is 0 . At a hexavalent weave vertex, we first draw a small circle C cutting out the hexavalent weave vertex and put a base point p at a generic position along C . Then we apply Lemma 3.14 and turn the problem to an intersection pairing computation along the interval C\setminus\{p\} . Without loss of generality, we may assume that the truncations of the weighted cycles look like the following. a_{1}\omega_{i} a^{\prime}_{1}\omega_{i} s_{i} a_{3}\omega_{i} a^{\prime}_{3}\omega_{i} s_{i} a_{5}\omega_{i} a^{\prime}_{5}\omega_{i} s_{i} a_{2}\omega_{j} a^{\prime}_{2}\omega_{j} s_{j} a_{4}\omega_{j} a^{\prime}_{4}\omega_{j} s_{j} a_{6}\omega_{j} a^{\prime}_{6}\omega_{j} s_{j} Figure 21. Truncation of weighted cycles near a hexavalent weave vertex; the black weighted chains are \eta^{\prime}_{1} and the purple weighted chains are \eta^{\prime}_{2} . Note that within each of the six collections of weighted chains near weave edges, the pairing between the chains in \eta^{\prime}_{1} and \eta^{\prime}_{2} is 0 . Therefore we only need to pay attention to how \eta^{\prime}_{1} in each collection is paired with \eta^{\prime}_{2} in every other collection. This allows us to do some homotopies to simplify: for example, to pair \eta^{\prime}_{1} in the 1 st collection with \eta^{\prime}_{2} in every other collection, we may simplify the configuration to the following, which yields a contribution of -\frac{1}{2}a_{1}(a^{\prime}_{2}-a^{\prime}_{3}-2a^{\prime}_{4}-a^{\prime}_{5}+a^{\prime}_{6})=a_{1}a^{\prime}_{1}. a_{1}\alpha_{i} s_{i} s_{j} s_{i} s_{j} s_{i} s_{j} a^{\prime}_{3}\alpha_{i} a^{\prime}_{5}\alpha_{i} a^{\prime}_{2}\alpha_{j} a^{\prime}_{4}\alpha_{j} a^{\prime}_{6}\alpha_{j} Figure 22. Simplifying the weighted chains. By similar computations, we will get \displaystyle\{\eta^{\prime}_{1},\eta^{\prime}_{2}\}_{C\setminus\{p\}}= \displaystyle a_{1}a^{\prime}_{1}+a_{2}(a^{\prime}_{4}+a^{\prime}_{5})+a_{3}(a^{\prime}_{2}-a^{\prime}_{4}+a^{\prime}_{6})-a_{4}(a^{\prime}_{1}-a^{\prime}_{3}+a^{\prime}_{5})-a_{5}(a^{\prime}_{2}+a^{\prime}_{3})-a_{6}a^{\prime}_{6} \displaystyle= \displaystyle a_{1}(a^{\prime}_{3}-a^{\prime}_{5})-a_{3}(a^{\prime}_{1}-a^{\prime}_{5})+a_{5}(a^{\prime}_{1}-a^{\prime}_{3}) \displaystyle= \displaystyle\det\begin{pmatrix}1&1&1\\ a_{1}&a_{3}&a_{5}\\ a^{\prime}_{1}&a^{\prime}_{3}&a^{\prime}_{5}\end{pmatrix}. For two external weave edges e and e^{\prime} incident to the same boundary interval, we first perform the following homotopy move (7) to the two weighted cycle representatives. e s_{i_{0}} e^{\prime} s_{i_{k}} \omega_{i_{0}} \omega_{i_{0}} \omega_{i_{k}} \omega_{i_{k}} s_{i_{1}} s_{i_{k-1}} \cdots \eta_{e} \eta_{e^{\prime}} = e s_{i_{0}} e^{\prime} s_{i_{k}} \alpha_{i_{0}} \alpha_{i_{k}} s_{i_{1}} s_{i_{k-1}} \cdots \eta_{e} \eta_{e^{\prime}} Figure 23. Homotoping weighted cycle representatives along a boundary interval. Then by definition, the intersection pairing \{\eta_{e},\eta_{e^{\prime}}\}=\frac{1}{2}(\alpha_{i_{0}},s_{i_{1}}\cdots s_{i_{k-1}}\alpha_{i_{k}}), which agrees with the definition of \{e,e^{\prime}\} in Definition 2.5 . ∎ Note that the weighted cycle representatives not only recover the intersection pairing between Y-cycles, they also allow us to define an intersection pairing between Y-cycles and weighted cycles. Definition 3.16 . Let \gamma be a Y-cycle and let \eta_{\gamma} be its weighted cycle representative. Then the intersection number between \gamma and an arbitrary weighted cycle \eta is defined to be \left\langle\gamma,\eta\right\rangle:=\{\eta_{\gamma},\eta\}. Since the intersection pairing between weighted cycles are invariant under homotopies of weighted cycles, we can hence deduce the following corollary. Corollary 3.17 . The intersection number between Y-cycles and weighted cycles is invariant under homotopies of weighted cycles. 3.5. Merodromies Recall that each weave defines a flag moduli space \mathcal{M}(\mathfrak{w}) , each point of which corresponds to a configuration of flags associated with faces of the weave \mathfrak{w} (Subsection 2.2 ). In this subsection, we define a map called merodromies , which allows us to view weighted cycles on \mathfrak{w} as functions on the flag moduli space \mathcal{M}(\mathfrak{w}) . In order to define the merodromies of weighted cycles, we need to first orient the weave edges, which corresponds to the choice of one of the h distances (Definition 2.10 ). In addition, we want to make sure that the weave edges are oriented in a compatible fashion so that the merodromies are homotopy-invariant. Definition 3.18 . A choice of orientations on all edges of a weave is said to be compatible if the following are satisfied: • each bivalent weave vertex is either a source or a sink; • there are two incoming edges and one outgoing edge at each trivalent weave vertex; • there are two adjacent incoming edges and two adjacent outgoing edges at each tetravalent weave vertex; • there are three adjacent incoming edges and three adjacent outgoing edges at each hexavalent weave vertex. \bullet \bullet Figure 24. Compatible orientation on weave edges. Definition 3.19 . Once a compatible orientation is chosen, we define the sign of an intersection between a weighted chain \eta and a weave edge as follows; note that the sign of an intersection is NOT the same as the intersection number defined in Definition 3.16 . + - Figure 25. Signs of the intersection between a weighted chain and a weave edge. Definition 3.20 . Given a configuration of flags in \mathcal{M}_{\mathrm{fr}}(\mathfrak{w}) , we first fix a decoration on each flag, making them decorated flags. Let \eta be a weighted chain. The initial decorated flag (resp. terminal decorated flag ) of \eta is the decorated flag along the boundary if the source (resp. target ) of \eta is on the boundary of the disk, and is otherwise the decorated flag associated with the face containing the interior endpoint. If \eta does not intersect any weave edges, then \eta must be contained inside some face, and hence the underlying undecorated flag of its initial and terminal decorated flags are the same. This implies that the terminal decoration is equal to t times the initial decoration for some t\in T , and we define the merodromy of \eta to be M^{\eta}:=t^{\mu}, where \mu is the weight associated with \eta . If \eta does intersect weave edges, we break \eta down into segments cut out by the weave edges. We associate the initial decorated flag with the segment containing the source of \eta and the terminal decorated flag with the segment containing the target of \eta , and associate the decorated flag of the local face for each of the remaining segments. Now suppose p is an intersection point between a weighted chain \eta with a weave edge e of color s_{i} such that s_{i}.\mu and \mu are the weights attached to \eta before and after \eta crossing e , and xN and yN are the decorated flags associated with the segments before and after \eta crossing e ; then the local merodromy M_{p} is defined to be M_{p}:=h_{\pm}(xN,yN)^{\mu}, where the subscript of h depends on the sign of the intersection at p . The merodromy along \eta is then defined to be M^{\eta}:=\prod_{p}M_{p} where p runs through all intersection points between \eta and weave edges. We extend merodromies for weighted cycles as follows: if a weighted cycle \eta consists of weighted chains \eta_{1},\dots,\eta_{k} , then its merodromy is M^{\eta}:=\prod_{i=1}^{k}M^{\eta_{i}}. Remark 3.21 . Note that by Lemma 2.11 , the h distances are invariant under the diagonal G -action; thus, (local) merodromies are not affected by the diagonal G -action. We will use this fact to simplify our proofs. Remark 3.22 . Note that the interior endpoints of weighted chains do not contribute anything to merodromies. Also, by convention, the empty weighted cycle has a merodromy of 1 . It may seem that the merodromy of a weighted cycle depends on the choice of decorations on flags. In the next few propositions, we will prove that merodromies are in fact independent of such choice and is also invariant under homotopies of weighted cycles. Proposition 3.23 . Suppose \eta is a weighted chain passing through a collection of weave edges e_{1},e_{2},\dots,e_{l} with colors s_{i_{1}},s_{i_{2}},\dots,s_{i_{l}} such that s_{i_{1}}s_{i_{2}}\cdots s_{i_{l}} form a reduced word for a Weyl group element w . Suppose all intersections are positive (resp. negative). Let x_{0}N,x_{1}N,\dots,x_{l}N be the decorated flags associated with the segments in \eta . Then M^{\eta}=h_{+}(x_{0}N,x_{l}N)^{\mu} (resp. M^{\eta}=h_{-}(x_{0}N,x_{l}N)^{\mu} where \mu is the weight attached to the end of \eta . Proof. Without loss of generality let us assume that all intersections are positive. Suppose x_{j-1}^{-1}x_{j}\in N\overline{s}_{i_{j}}h_{j}N for j=1,2,\dots,l . Then on the one hand, by definition, M^{\eta}=\prod_{j=1}^{l}h_{j}^{w_{>j}.\mu}=\left(\prod_{j=1}^{l}w_{>j}^{-1}(h_{j})\right)^{\mu}, where w_{>j}:=s_{i_{j+1}}s_{i_{j+2}}\cdots s_{i_{l}} . On the other hand, x_{0}^{-1}x_{l}=(x_{0}^{-1}x_{1})(x_{1}^{-1}x_{2})\cdots(x_{l-1}^{-1}x_{l})\in(N\overline{s}_{i_{1}}h_{1}N)(N\overline{s}_{i_{2}}h_{2}N)\cdots(N\overline{s}_{i_{l}}h_{l}N)=N\overline{w}\prod_{j=1}^{l}w_{>j}^{-1}(h_{j})N. Thus h_{+}(x_{0}N,x_{l}N)=\prod_{j=1}^{l}w_{>j}^{-1}(h_{j}) as well. ∎ Proposition 3.24 . If F is a face where a weighted chain \eta passes through in the middle, i.e., it is neither the first face nor the last face, then merodromy M^{\eta} only depends on the undecorated flag associated with F and is independent of the decoration. Proof. Suppose the two weave edges before and after the face F are of colors s_{i} and s_{j} and suppose \eta passes through faces with with decorated flags x_{0}N , x_{1}N , and x_{2}N (see Figure 26 ). Changing the decoration at F is the same as changing the decorated flag x_{1}N to x_{1}tN for some t\in T . Under such a change, we have h_{\pm}(x_{0}N,x_{1}tN)=th_{\pm}(x_{0}N,x_{1}N) and h_{\pm}(x_{1}tN,x_{2}N)=s_{j}(t^{-1})h_{\pm}(x_{1}N,x_{2}N) , where the signs in the subscript depend on the orientations of the weave edges. Suppose the weight associated with the middle portion of \eta is \mu . Then the local contribution to the merodromy M^{\eta} is \displaystyle h_{\pm}(x_{0}N,x_{1}tN)^{\mu}\cdot h_{\pm}(x_{1}tN,x_{2}N)^{s_{j}.\mu}= \displaystyle(th_{\pm}(x_{0}N,x_{1}N))^{\mu}\cdot(s_{j}(t^{-1})h_{\pm}(x_{1}N,x_{2}N))^{s_{j}.\mu} \displaystyle= \displaystyle t^{\mu}\cdot h_{\pm}(x_{0}N,x_{1}N)^{\mu}\cdot t^{-\mu}\cdot h_{\pm}(x_{1}N,x_{2}N)^{s_{j}.\mu} \displaystyle= \displaystyle h_{\pm}(x_{0}N,x_{1}N)^{\mu}\cdot h_{\pm}(x_{1}N,x_{2}N)^{s_{j}.\mu}. This computation shows that the local contribution before and after the change of decoration are equal, and hence M^{\eta} is invariant under the change of decorations on the intermediate faces. ∎ s_{i} s_{j} x_{1}N x_{2}N x_{0}N F s_{i}.\mu \mu s_{j}.\mu Figure 26. A weighted chain passing through a face. Proposition 3.25 . Suppose F is a face that contains an interior endpoint p for weighted chains \eta_{1},\eta_{2},\dots,\eta_{k} , each is oriented towards p with weights \mu_{1},\mu_{2},\dots,\mu_{k} , respectively, such that \mu_{1}+\mu_{2}+\cdots+\mu_{k}=0 . Then the total merodromy \prod_{i=1}^{k}M^{\eta_{k}} only depends on the undecorated flag associated with F and is independent of the decoration. Proof. Without loss of generality we may assume that the decorated flag associated with the face F is N . Now if we change it to tN for some t\in T , the combined change to the total merodromy is \prod_{i=1}^{k}t^{\mu_{i}}=t^{\sum_{i=1}^{k}\mu_{i}}=t^{0}=1.\qed Proposition 3.26 . Merodromies are invariant under homotopies of weighted cycles. Proof. Since all contributions to merodromies are local, we just need to go through the list of homotopy moves (see Figures 8 and 11 ) and prove the invariance. Homotopy move (0) can be achieved by first adding an interior endpoint (homotopy move (8)), then pushing this interior endpoint through the weave edge (homotopy move (12)), and then deleting the interior endpoint (homotopy move (8)) again. Thus, its homotopy invariance follows from the homotopy invariance of other moves. For the homotopy move (1), let us assume without loss of generality that the weave edge is oriented downward. Let us suppose the decorated flags are xN on the left and yN on the right with h_{+}(xN,yN)=h , i.e., x^{-1}y\in N\overline{s}_{i}hN . Then the merodromy on the LHS is h^{s_{i}.\mu} . Note that y^{-1}x\in Nh^{-1}\overline{s}_{i}^{-1}N=Nh^{-1}\overline{\overline{s}}_{i}N=N\overline{\overline{s}}_{i}(s_{i}(h^{-1}))N. Thus, the merodromy on the RHS is (s_{i}(h^{-1}))^{-\mu}=h^{s_{i}.\mu} as well. For the homotopy move (2), the initial decorated flag and terminal decorated flag of the weighted chain are the same and hence M^{\eta}=1 . For the homotopy move (3), we observe that it is almost the composition of homotopy moves (0) and (2), except that one of the two merodromies is computed with the opposite h distance (i.e., h_{\mp} instead of h_{\pm} ). It then follows from Lemma 2.12 that this homotopy move results in a sign ((-1)^{\alpha_{i}^{\vee}})^{\mu}=(-1)^{\left\langle\alpha_{i}^{\vee},\mu\right\rangle} . For the homotopy move (4), since the picture has a rotational symmetry, we may assume without loss of generality that all the weave edges are oriented downward. Furthermore, we may first use the diagonal G -action to move the decorated flags so that the decorated flag associated with the left face is N ; we then apply Proposition 3.24 to assume that the decorated flag associated with the top face is of the form x_{i}(p)\overline{s}_{i}N ; under this setup, the merodromy across the left weave edge is 1 . The decorated flag associated with the right face is of the form x_{i}(q)\overline{s}_{i}hN for some h\in T and q\neq p , and as a result, the merodromy across the bottom weave edge is h_{-}(x_{i}(q)\overline{s}_{i}hN,N)^{\mu}=s_{i}(h^{-1})^{\mu}=h^{-\mu} (note that s_{i}.\mu=\mu by assumption). The remaining merodromy across the right weave edge is h_{+}(x_{i}(p)\overline{s}_{i}N,x_{i}(q)\overline{s}_{i}hN)^{\mu} . By applying the identity \overline{s}_{i}^{-1}x_{i}(a)=x_{i}(-a^{-1})(-a)^{-\alpha_{i}^{\vee}}y_{i}(a^{-1}), we note that N\overline{s}_{i}^{-1}x_{i}(p)^{-1}x_{i}(q)\overline{s}_{i}hN=N\overline{s}_{i}^{-1}x_{i}(q-p)\overline{s}_{i}hN=N(p-q)^{-\alpha_{i}^{\vee}}\overline{s}_{i}hN=N\overline{s}_{i}(p-q)^{\alpha_{i}^{\vee}}hN. Since s_{i}.\mu=\mu , we know that \left\langle\alpha_{i}^{\vee},\mu\right\rangle=0 . Thus, we can conclude that the merodromy across the right weave edge is ((p-q)^{\alpha_{i}^{\vee}}h)^{\mu}=h^{\mu}. Thus, the total merodromy is 1\cdot h^{-\mu}\cdot h^{\mu}=1 . For the homotopy moves (5) and (6), we can add bivalent interior endpoints on the left-most and the right-most points of the circle to break the circle into two semicircular arcs. Note that by definition, reversing the orientation on a weighted chain without changing the weights results in a merodromy that is the inverse of the original. Thus, the invariance of merodromies for homotopy moves (5) and (6) is equivalent to showing that the merodromy across the upper semicircular arc is equal to the merodromy across the lower semicircular arc, which follows straight from Proposition 3.23 . For the homotopy move (7), it suffices to notice that the merodromy from the intersection on the RHS is 1 due to the condition on the decorated flags along the boundary (Definition 2.13 ). Homotopy moves (8)-(11) do not involve any non-trivial contribution to merodromies and there is nothing to show. Lastly, for the homotopy move (12), let us assume without loss of generality that the weave edge is oriented downward; the other orientation can be verified similarly with the h distance h_{\pm} replaced by h_{\mp} . Let xN be the decorated flag on the left of the weave edge and let yN be the decorated flag on the right of the weave edge. Suppose x^{-1}y\in N\overline{s}_{i}hN . Then on the one hand, the merodromy on the LHS is h^{\mu_{1}} . On the other hand, since y^{-1}x\in Nh^{-1}\overline{s}_{i}^{-1}N=N\overline{\overline{s}}_{i}s_{i}(h^{-1})N , the merodromy on the RHS is (s_{i}(h^{-1}))^{s_{i}.\mu_{2}+\cdots+s_{i}.\mu_{k}}=h^{-\mu_{2}-\cdots-\mu_{k}}=h^{\mu_{1}}.\qed As a result of Proposition 3.26 , we can view merodromies as an algebra homomorphism M^{\bullet}:\mathcal{W}(\mathfrak{w})\longrightarrow\mathcal{O}(\mathcal{M}_{\mathrm{fr}}(\mathfrak{w})). Recall that within the weighted cycle algebra, there is a subalgebra \mathcal{W}_{\mathrm{abs}}(\mathfrak{w}) generated by weighted absolute cycles. From the projection map \mathcal{M}_{\mathrm{fr}}(\mathfrak{w})\rightarrow\mathcal{M}(\mathfrak{w}) , we also get a subalgebra \mathcal{O}(\mathcal{M}(\mathfrak{w}))\subset\mathcal{O}(\mathcal{M}_{\mathrm{fr}}(\mathfrak{w})) . Corollary 3.27 . The merodromy homomorphism M^{\bullet}:\mathcal{W}(\mathfrak{w})\rightarrow\mathcal{O}(\mathcal{M}_{\mathrm{fr}}(\mathfrak{w})) restricts to a homomorphism M^{\bullet}:\mathcal{W}_{\mathrm{abs}}(\mathfrak{w})\longrightarrow\mathcal{O}(\mathcal{M}(\mathfrak{w})). Proof. Since weighted absolute cycles do not have endpoints along the boundary of the disk, the framing data of decorated flags along the boundary of the disk do not affect the merodromies along weighted absolute cycles. This shows that the merodromy homomorphism does restrict to a homomorphism from \mathcal{W}_{\mathrm{abs}}(\mathfrak{w}) to \mathcal{O}(\mathcal{M}(\mathfrak{w})) . ∎ Note that by construction, merodromies along weighted cycles depend on the choice of compatible orientation on the weave edges. Next, let us discuss how different choices of compatible orientations may affect merodromies. Definition 3.28 . A dissecting path on a weave \mathfrak{w} is a simple curve consisting of consecutive weave edges such that • at a tetravalent weave vertex on its path, the two weave edges are the i th and the (i+2) nd edges (modulo 4 ) with respect to any cyclic orientation on the incident weave edges; • at a hexavalent weave vertex on its path, the two weave edges are the i th and the (i+3) rd edges (modulo 6 ) with respect to any cyclic orientation on the incident weave edges; • it is maximal with respect to inclusion. Note that a dissecting path either forms a closed loop or travels across the disk, and hence it cuts the disk into two connected components. Lemma 3.29 . Any two choices of compatible orientations on the same weave differ by reversing orientations on a collection of dissecting paths. Proof. Let us do an induction on the number n of weave edges along which the two choices of compatible orientations disagree. There is nothing to show for the base case n=0 . Now let e be a weave edge along which the two choices disagree. Then at each of the endpoints of e , there must be another edge where the two choices disagree; in particular, if the endpoint is a tetravalent or a hexavalent weave vertex, the weave edge opposite to e must have disagreeing orientations as well. Thus we can extend e into a path consisting of consecutive weave edges. Eventually this process has to stop, either when the path closes itself up into a loop or goes to the boundary of the disk, and thus giving us a dissecting path. Now reversing one choice of compatible orientations along this dissecting path gives us another choice of compatible orientations with fewer disagreeing edges compared to the other choice, and the proof is complete by induction. ∎ Proposition 3.30 . Let us fix a choice of compatible orientations on \mathfrak{w} . Let \eta be a weighted cycle with merodromy M_{1}^{\eta} . Let \gamma be a dissecting path and let M_{2}^{\eta} be the merodromy along \eta with respect to the choice of compatible orientations obtained by reversing the edges in \gamma . Then M_{2}^{\eta}=(-1)^{\left\langle\gamma,\eta\right\rangle}M_{1}^{\eta} . Proof. By Lemma 2.12 , the two h -distances between two decorated flags separated by a weave edge of color s_{i} differ by (-1)^{\alpha_{i}^{\vee}} . Therefore we may conclude that M_{2}^{\eta}=M_{1}^{\eta}\prod_{k}(-1)^{\left\langle\alpha_{i_{k}}^{\vee},\mu_{k}\right\rangle} , where k runs through all intersection points between \eta and \gamma , \mu_{k} is the weight on \eta right after the intersection point k , and s_{i_{k}} is the color of the weave edge at the intersection point k . The proposition then follows from the fact that \sum_{k}\left\langle\alpha_{i_{k}}^{\vee},\mu_{k}\right\rangle is precisely the intersection number \left\langle\gamma,\eta\right\rangle . ∎ Example 3.31 . Below is an example of two choices of compatible orientations on the same \mathrm{SL}_{3} -weave together with the same weighted absolute cycle whose merodromies would differ by a factor of -1 . s_{1}.\omega_{1} \omega_{1} \omega_{1} s_{1}.\omega_{1} w_{0}.\omega_{1} s_{1}.\omega_{1} \omega_{1} \omega_{1} s_{1}.\omega_{1} \omega_{1} \omega_{1} s_{1}.\omega_{1} w_{0}.\omega_{1} s_{1}.\omega_{1} \omega_{1} \omega_{1} Figure 27. Two choices of compatible orientations such that the merodromies along the same weighted cycle differ by a sign. Remark 3.32 . Let us discuss the significance of introducing bivalent weave vertex for computing merodromies. First, let us observe that without bivalent weave vertices, there exist weaves that do not admit any choices of compatible orientations (e.g., the left picture in Figure 28 ). In addition, although weave equivalences are local moves (Definition 2.7 ), there is no guarantee that one can adjust a compatible choice of orientations locally to obtain a compatible choice of orientations after a weave equivalence (e.g., the right picture in Figure 28 ). {\longleftrightarrow} Figure 28. Examples of non-existence of compatible orientations. By introducing bivalent weave vertices, we can give these weaves compatible orientations. \bullet {\longleftrightarrow} \bullet \bullet Figure 29. Compatible orientations after adding bivalent weave vertices. Note that bivalent weave vertices are not needed if we are not computing merodromies. Also, since bivalent weave vertices only introduce a possible sign to merodromies, they are not needed if the base field is of characteristic 2 . 3.6. Mutations In this subsection we describe how a family of weighted cycles change under a weave mutation at a short I-cycle. Definition 3.33 . Suppose \mathfrak{w} and \mathfrak{w}^{\prime} are two weaves differing by a mutation at a short I-cycle at the weave edge e . If a weighted chain \eta on \mathfrak{w} has an intersection number \left\langle e,\eta\right\rangle=0 , then we can use homotopy move (4) to move it out of the local mutating region and keep it as a weighted chain on \mathfrak{w}^{\prime} . In contrast, if \eta is a weighted chain with an intersection number \left\langle e,\eta\right\rangle=1 , then we define the mutation of \eta at e to be \eta_{1}+\eta_{2} in \mathcal{W}(\mathfrak{w}^{\prime}) , where \eta_{1} and \eta_{2} are depicted in Figure 30 . s_{i}.\mu \mu \eta \longrightarrow s_{i}.\mu \mu s_{i}.\mu \mu \eta_{1} + s_{i}.\mu \mu s_{i}.\mu \mu \eta_{2} Figure 30. Mutation of a weighted chain whose intersection number with the short I-cycle is 1 . Definition 3.34 . Let e be a short I-cycle on a weave \mathfrak{w} . Within the weighted cycle algebra \mathcal{W}(\mathfrak{w}) , we define the subalgebra \mathcal{W}_{e}^{+}(\mathfrak{w}) of non-negative weighted cycles at e to be the subalgebra generated by weighted cycles with representatives such that all of whose weighted chains have non-negative intersection numbers with e . Then the mutation operation above induces an algebra homomorphism \mu_{e}:\mathcal{W}_{e}^{+}(\mathfrak{w})\longrightarrow\mathcal{W}(\mathfrak{w}^{\prime}), where \mathfrak{w}^{\prime} is the weave obtained from \mathfrak{w} by a mutation at e . Theorem 3.35 . The mutation homomorphism \mu_{e}:\mathcal{W}_{e}^{+}(\mathfrak{w})\longrightarrow\mathcal{W}(\mathfrak{w}^{\prime}) commutes with the merodromy homomorphisms. Proof. It suffices to prove that when \left\langle e,\eta\right\rangle=1 , M^{\eta}=M^{\eta_{1}}+M^{\eta_{2}} where \eta_{1}+\eta_{2} is the mutation of \eta at e . Let us use the weights in Figure 30 for reference. By Definition 3.16 , we know that \left\langle\alpha_{i}^{\vee},\mu\right\rangle=1 . Without loss of generality, let us assume that the orientations on weave edges and the surrounding decorated flags are chosen as follows (with p\neq q and p,q\neq 0 ). It follows that M^{\eta}=h_{2}^{\mu} . \hbox to169.93pt{\vbox to71.21pt{\pgfpicture\makeatletter\hbox{\hskip 85.01631pt\lower-35.20242pt\hbox to0.0pt{\lxSVG@begingroup@{_scopebegin=1} \lxSVG@begingroup@{stroke=#000000} \lxSVG@begingroup@{fill=#000000} \lxSVG@setlinewidth{\the\pgflinewidth}\lxSVG@begingroup@{stroke-width=0.4pt} \lx@inpgf@ignorespaces\nullfont\hbox to0.0pt{\lxSVG@begingroup@{_scopebegin=1} {}{{}}{} {}{}\lxSVG@begingroup@{_scopebegin=1} \color[rgb]{0,0,1}{}\lxSVG@stroke\lxSVG@drawpath@unclipped{M -39.37 -39.37 L -19.69 0}{fill:none} \lx@inpgf@ignorespaces{\lxSVG@begingroup@{_scopebegin=1} {}{{}{}\lx@inpgf@ignorespaces}{}{}{}\lx@inpgf@ignorespaces{{}}{{}}{{}{}\lx@inpgf@ignorespaces}{{}{}\lx@inpgf@ignorespaces}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}\lxSVG@begingroup@{_scopebegin=1} {\lxSVG@begingroup@{_scopebegin=1} {{\lx@inpgf@ignorespaces}} {{}{{{\lx@inpgf@ignorespaces}}{{{\lx@inpgf@ignorespaces}}{\lxSVG@begingroup@{_scopebegin=1} \lxSVG@transformcm{0.4472}{0.89442}{-0.89442}{0.4472}{-21.42883pt}{-14.40465pt}\lxSVG@begingroup@{transform=matrix(0.4472 0.89442 -0.89442 0.4472 -29.65 -19.93)} \lxSVG@begingroup@{_scopebegin=1} \lxSVG@begingroup@{stroke-dasharray=none,stroke-dashoffset=0.0pt} \lxSVG@begingroup@{stroke-linecap=round} \lxSVG@begingroup@{stroke-linejoin=round} \lxSVG@drawpath@unclipped{M -2.88 3.32 C -2.35 1.33 -1.18 0.39 0 0 C -1.18 -0.39 -2.35 -1.33 -2.88 -3.32}{fill:none} \lxSVG@closescope \lxSVG@closescope }}{{\lx@inpgf@ignorespaces}}}} \lxSVG@closescope }\lxSVG@closescope }{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}} {{{}}} }{{}{}\lx@inpgf@ignorespaces}{{}{}\lx@inpgf@ignorespaces}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}} {{{}}} } \lx@inpgf@ignorespaces \lxSVG@closescope } \lxSVG@closescope {}{{}}{} {}{}\lxSVG@begingroup@{_scopebegin=1} \color[rgb]{0,0,1}{}\lxSVG@stroke\lxSVG@drawpath@unclipped{M -39.37 39.37 L -19.69 0}{fill:none} \lx@inpgf@ignorespaces{\lxSVG@begingroup@{_scopebegin=1} {}{{}{}\lx@inpgf@ignorespaces}{}{}{}\lx@inpgf@ignorespaces{{}}{{}}{{}{}\lx@inpgf@ignorespaces}{{}{}\lx@inpgf@ignorespaces}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}\lxSVG@begingroup@{_scopebegin=1} {\lxSVG@begingroup@{_scopebegin=1} {{\lx@inpgf@ignorespaces}} {{}{{{\lx@inpgf@ignorespaces}}{{{\lx@inpgf@ignorespaces}}{\lxSVG@begingroup@{_scopebegin=1} \lxSVG@transformcm{0.4472}{-0.89442}{0.89442}{0.4472}{-21.42883pt}{14.40465pt}\lxSVG@begingroup@{transform=matrix(0.4472 -0.89442 0.89442 0.4472 -29.65 19.93)} \lxSVG@begingroup@{_scopebegin=1} \lxSVG@begingroup@{stroke-dasharray=none,stroke-dashoffset=0.0pt} \lxSVG@begingroup@{stroke-linecap=round} \lxSVG@begingroup@{stroke-linejoin=round} \lxSVG@drawpath@unclipped{M -2.88 3.32 C -2.35 1.33 -1.18 0.39 0 0 C -1.18 -0.39 -2.35 -1.33 -2.88 -3.32}{fill:none} \lxSVG@closescope \lxSVG@closescope }}{{\lx@inpgf@ignorespaces}}}} \lxSVG@closescope }\lxSVG@closescope }{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}} {{{}}} }{{}{}\lx@inpgf@ignorespaces}{{}{}\lx@inpgf@ignorespaces}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}} {{{}}} } \lx@inpgf@ignorespaces \lxSVG@closescope } \lxSVG@closescope {}{{}}{} {}{}\lxSVG@begingroup@{_scopebegin=1} \color[rgb]{0,0,1}{}\lxSVG@stroke\lxSVG@drawpath@unclipped{M 39.37 -39.37 L 19.69 0}{fill:none} \lx@inpgf@ignorespaces{\lxSVG@begingroup@{_scopebegin=1} {}{{}{}\lx@inpgf@ignorespaces}{}{}{}\lx@inpgf@ignorespaces{{}}{{}}{{}{}\lx@inpgf@ignorespaces}{{}{}\lx@inpgf@ignorespaces}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}\lxSVG@begingroup@{_scopebegin=1} {\lxSVG@begingroup@{_scopebegin=1} {{\lx@inpgf@ignorespaces}} {{}{{{\lx@inpgf@ignorespaces}}{{{\lx@inpgf@ignorespaces}}{\lxSVG@begingroup@{_scopebegin=1} \lxSVG@transformcm{-0.4472}{0.89442}{-0.89442}{-0.4472}{21.42883pt}{-14.40465pt}\lxSVG@begingroup@{transform=matrix(-0.4472 0.89442 -0.89442 -0.4472 29.65 -19.93)} \lxSVG@begingroup@{_scopebegin=1} \lxSVG@begingroup@{stroke-dasharray=none,stroke-dashoffset=0.0pt} \lxSVG@begingroup@{stroke-linecap=round} \lxSVG@begingroup@{stroke-linejoin=round} \lxSVG@drawpath@unclipped{M -2.88 3.32 C -2.35 1.33 -1.18 0.39 0 0 C -1.18 -0.39 -2.35 -1.33 -2.88 -3.32}{fill:none} \lxSVG@closescope \lxSVG@closescope }}{{\lx@inpgf@ignorespaces}}}} \lxSVG@closescope }\lxSVG@closescope }{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}} {{{}}} }{{}{}\lx@inpgf@ignorespaces}{{}{}\lx@inpgf@ignorespaces}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}} {{{}}} } \lx@inpgf@ignorespaces \lxSVG@closescope } \lxSVG@closescope {}{{}}{} {}{}\lxSVG@begingroup@{_scopebegin=1} \color[rgb]{0,0,1}{}\lxSVG@stroke\lxSVG@drawpath@unclipped{M 19.69 0 L 39.37 39.37}{fill:none} \lx@inpgf@ignorespaces{\lxSVG@begingroup@{_scopebegin=1} {}{{}{}\lx@inpgf@ignorespaces}{}{}{}\lx@inpgf@ignorespaces{{}}{{}}{{}{}\lx@inpgf@ignorespaces}{{}{}\lx@inpgf@ignorespaces}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}\lxSVG@begingroup@{_scopebegin=1} {\lxSVG@begingroup@{_scopebegin=1} {{\lx@inpgf@ignorespaces}} {{}{{{\lx@inpgf@ignorespaces}}{{{\lx@inpgf@ignorespaces}}{\lxSVG@begingroup@{_scopebegin=1} \lxSVG@transformcm{0.4472}{0.89442}{-0.89442}{0.4472}{21.2503pt}{14.04811pt}\lxSVG@begingroup@{transform=matrix(0.4472 0.89442 -0.89442 0.4472 29.4 19.44)} \lxSVG@begingroup@{_scopebegin=1} \lxSVG@begingroup@{stroke-dasharray=none,stroke-dashoffset=0.0pt} \lxSVG@begingroup@{stroke-linecap=round} \lxSVG@begingroup@{stroke-linejoin=round} \lxSVG@drawpath@unclipped{M -2.88 3.32 C -2.35 1.33 -1.18 0.39 0 0 C -1.18 -0.39 -2.35 -1.33 -2.88 -3.32}{fill:none} \lxSVG@closescope \lxSVG@closescope }}{{\lx@inpgf@ignorespaces}}}} \lxSVG@closescope }\lxSVG@closescope }{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}} {{{}}} }{{}{}\lx@inpgf@ignorespaces}{{}{}\lx@inpgf@ignorespaces}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}} {{{}}} } \lx@inpgf@ignorespaces \lxSVG@closescope } \lxSVG@closescope {}{{}}{} {}{}\lxSVG@begingroup@{_scopebegin=1} \color[rgb]{0,0,1}{}\lxSVG@stroke\lxSVG@drawpath@unclipped{M -19.69 0 L 19.69 0}{fill:none} \lx@inpgf@ignorespaces{\lxSVG@begingroup@{_scopebegin=1} {}{{}{}\lx@inpgf@ignorespaces}{}{}{}\lx@inpgf@ignorespaces{{}}{{}}{{}{}\lx@inpgf@ignorespaces}{{}{}\lx@inpgf@ignorespaces}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}\lxSVG@begingroup@{_scopebegin=1} {\lxSVG@begingroup@{_scopebegin=1} {{\lx@inpgf@ignorespaces}} {{}{{{\lx@inpgf@ignorespaces}}{{{\lx@inpgf@ignorespaces}}{\lxSVG@begingroup@{_scopebegin=1} \lxSVG@transformcm{1.0}{0.0}{0.0}{1.0}{-0.20001pt}{0.0pt}\lxSVG@begingroup@{transform=matrix(1.0 0.0 0.0 1.0 -0.28 0)} \lxSVG@begingroup@{_scopebegin=1} \lxSVG@begingroup@{stroke-dasharray=none,stroke-dashoffset=0.0pt} \lxSVG@begingroup@{stroke-linecap=round} \lxSVG@begingroup@{stroke-linejoin=round} \lxSVG@drawpath@unclipped{M -2.88 3.32 C -2.35 1.33 -1.18 0.39 0 0 C -1.18 -0.39 -2.35 -1.33 -2.88 -3.32}{fill:none} \lxSVG@closescope \lxSVG@closescope }}{{\lx@inpgf@ignorespaces}}}} \lxSVG@closescope }\lxSVG@closescope }{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}} {{}} }{{}{}\lx@inpgf@ignorespaces}{{}{}\lx@inpgf@ignorespaces}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}} {{}} } \lx@inpgf@ignorespaces \lxSVG@closescope } \lxSVG@closescope {{}}\lx@inpgf@ignorespaces\hbox{\hbox{{\lxSVG@begingroup@{_scopebegin=1} {{}{}{{ {}{}}}{ {}{}} {{}{{\lx@inpgf@ignorespaces}}}{{}{\lx@inpgf@ignorespaces}}{}{{}{\lx@inpgf@ignorespaces}} {\lx@inpgf@ignorespaces }{{{{\lx@inpgf@ignorespaces}}\lxSVG@begingroup@{_scopebegin=1} \lxSVG@transformcm{1.0}{0.0}{0.0}{1.0}{-4.56248pt}{-31.86942pt}\lxSVG@begingroup@{transform=matrix(1.0 0.0 0.0 1.0 -6.31 -44.1)} \pgfsys@hbox{58}\lxSVG@closescope }}} \lxSVG@closescope }}} {{}}\lx@inpgf@ignorespaces\hbox{\hbox{{\lxSVG@begingroup@{_scopebegin=1} {{}{}{{ {}{}}}{ {}{}} {{}{{\lx@inpgf@ignorespaces}}}{{}{\lx@inpgf@ignorespaces}}{}{{}{\lx@inpgf@ignorespaces}} {\lx@inpgf@ignorespaces }{{{{\lx@inpgf@ignorespaces}}\lxSVG@begingroup@{_scopebegin=1} \lxSVG@transformcm{1.0}{0.0}{0.0}{1.0}{-13.85098pt}{25.73055pt}\lxSVG@begingroup@{transform=matrix(1.0 0.0 0.0 1.0 -19.17 35.6)} \pgfsys@hbox{58}\lxSVG@closescope }}} \lxSVG@closescope }}} {{}}\lx@inpgf@ignorespaces\hbox{\hbox{{\lxSVG@begingroup@{_scopebegin=1} {{}{}{{ {}{}}}{ {}{}} {{}{{\lx@inpgf@ignorespaces}}}{{}{\lx@inpgf@ignorespaces}}{}{{}{\lx@inpgf@ignorespaces}} {\lx@inpgf@ignorespaces }{{{{\lx@inpgf@ignorespaces}}\lxSVG@begingroup@{_scopebegin=1} \lxSVG@transformcm{1.0}{0.0}{0.0}{1.0}{-81.6833pt}{-2.5pt}\lxSVG@begingroup@{transform=matrix(1.0 0.0 0.0 1.0 -113.03 -3.46)} \pgfsys@hbox{58}\lxSVG@closescope }}} \lxSVG@closescope }}} {{}}\lx@inpgf@ignorespaces\hbox{\hbox{{\lxSVG@begingroup@{_scopebegin=1} {{}{}{{ {}{}}}{ {}{}} {{}{{\lx@inpgf@ignorespaces}}}{{}{\lx@inpgf@ignorespaces}}{}{{}{\lx@inpgf@ignorespaces}} {\lx@inpgf@ignorespaces }{{{{\lx@inpgf@ignorespaces}}\lxSVG@begingroup@{_scopebegin=1} \lxSVG@transformcm{1.0}{0.0}{0.0}{1.0}{32.2319pt}{-2.5pt}\lxSVG@begingroup@{transform=matrix(1.0 0.0 0.0 1.0 44.6 -3.46)} \pgfsys@hbox{58}\lxSVG@closescope }}} \lxSVG@closescope }}} \lxSVG@closescope {\lx@inpgf@ignorespaces}{\lx@inpgf@ignorespaces}{\lx@inpgf@ignorespaces}\hss}\lxSVG@discardpath\lxSVG@closescope \hss}}\lxSVG@closescope\endpgfpicture}}\hskip 28.45274pt\longleftrightarrow\hskip 28.45274pt\hbox to169.93pt{\vbox to71.21pt{\pgfpicture\makeatletter\hbox{\hskip 85.01631pt\lower-35.20242pt\hbox to0.0pt{\lxSVG@begingroup@{_scopebegin=1} \lxSVG@begingroup@{stroke=#000000} \lxSVG@begingroup@{fill=#000000} \lxSVG@setlinewidth{\the\pgflinewidth}\lxSVG@begingroup@{stroke-width=0.4pt} \lx@inpgf@ignorespaces\nullfont\hbox to0.0pt{\lxSVG@begingroup@{_scopebegin=1} {}{{}}{} {}{}\lxSVG@begingroup@{_scopebegin=1} \color[rgb]{0,0,1}{}\lxSVG@stroke\lxSVG@drawpath@unclipped{M -39.37 -39.37 L 0 -19.69}{fill:none} \lx@inpgf@ignorespaces{\lxSVG@begingroup@{_scopebegin=1} {}{{}{}\lx@inpgf@ignorespaces}{}{}{}\lx@inpgf@ignorespaces{{}}{{}}{{}{}\lx@inpgf@ignorespaces}{{}{}\lx@inpgf@ignorespaces}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}\lxSVG@begingroup@{_scopebegin=1} {\lxSVG@begingroup@{_scopebegin=1} {{\lx@inpgf@ignorespaces}} {{}{{{\lx@inpgf@ignorespaces}}{{{\lx@inpgf@ignorespaces}}{\lxSVG@begingroup@{_scopebegin=1} \lxSVG@transformcm{0.89442}{0.4472}{-0.4472}{0.89442}{-14.40465pt}{-21.42883pt}\lxSVG@begingroup@{transform=matrix(0.89442 0.4472 -0.4472 0.89442 -19.93 -29.65)} \lxSVG@begingroup@{_scopebegin=1} \lxSVG@begingroup@{stroke-dasharray=none,stroke-dashoffset=0.0pt} \lxSVG@begingroup@{stroke-linecap=round} \lxSVG@begingroup@{stroke-linejoin=round} \lxSVG@drawpath@unclipped{M -2.88 3.32 C -2.35 1.33 -1.18 0.39 0 0 C -1.18 -0.39 -2.35 -1.33 -2.88 -3.32}{fill:none} \lxSVG@closescope \lxSVG@closescope }}{{\lx@inpgf@ignorespaces}}}} \lxSVG@closescope }\lxSVG@closescope }{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}} {{{}}} }{{}{}\lx@inpgf@ignorespaces}{{}{}\lx@inpgf@ignorespaces}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}} {{{}}} } \lx@inpgf@ignorespaces \lxSVG@closescope } \lxSVG@closescope {}{{}}{} {}{}\lxSVG@begingroup@{_scopebegin=1} \color[rgb]{0,0,1}{}\lxSVG@stroke\lxSVG@drawpath@unclipped{M 39.37 -39.37 L 0 -19.69}{fill:none} \lx@inpgf@ignorespaces{\lxSVG@begingroup@{_scopebegin=1} {}{{}{}\lx@inpgf@ignorespaces}{}{}{}\lx@inpgf@ignorespaces{{}}{{}}{{}{}\lx@inpgf@ignorespaces}{{}{}\lx@inpgf@ignorespaces}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}\lxSVG@begingroup@{_scopebegin=1} {\lxSVG@begingroup@{_scopebegin=1} {{\lx@inpgf@ignorespaces}} {{}{{{\lx@inpgf@ignorespaces}}{{{\lx@inpgf@ignorespaces}}{\lxSVG@begingroup@{_scopebegin=1} \lxSVG@transformcm{-0.89442}{0.4472}{-0.4472}{-0.89442}{14.40465pt}{-21.42883pt}\lxSVG@begingroup@{transform=matrix(-0.89442 0.4472 -0.4472 -0.89442 19.93 -29.65)} \lxSVG@begingroup@{_scopebegin=1} \lxSVG@begingroup@{stroke-dasharray=none,stroke-dashoffset=0.0pt} \lxSVG@begingroup@{stroke-linecap=round} \lxSVG@begingroup@{stroke-linejoin=round} \lxSVG@drawpath@unclipped{M -2.88 3.32 C -2.35 1.33 -1.18 0.39 0 0 C -1.18 -0.39 -2.35 -1.33 -2.88 -3.32}{fill:none} \lxSVG@closescope \lxSVG@closescope }}{{\lx@inpgf@ignorespaces}}}} \lxSVG@closescope }\lxSVG@closescope }{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}} {{{}}} }{{}{}\lx@inpgf@ignorespaces}{{}{}\lx@inpgf@ignorespaces}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}} {{{}}} } \lx@inpgf@ignorespaces \lxSVG@closescope } \lxSVG@closescope {}{{}}{} {}{}\lxSVG@begingroup@{_scopebegin=1} \color[rgb]{0,0,1}{}\lxSVG@stroke\lxSVG@drawpath@unclipped{M -39.37 39.37 L 0 19.69}{fill:none} \lx@inpgf@ignorespaces{\lxSVG@begingroup@{_scopebegin=1} {}{{}{}\lx@inpgf@ignorespaces}{}{}{}\lx@inpgf@ignorespaces{{}}{{}}{{}{}\lx@inpgf@ignorespaces}{{}{}\lx@inpgf@ignorespaces}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}\lxSVG@begingroup@{_scopebegin=1} {\lxSVG@begingroup@{_scopebegin=1} {{\lx@inpgf@ignorespaces}} {{}{{{\lx@inpgf@ignorespaces}}{{{\lx@inpgf@ignorespaces}}{\lxSVG@begingroup@{_scopebegin=1} \lxSVG@transformcm{0.89442}{-0.4472}{0.4472}{0.89442}{-14.40465pt}{21.42883pt}\lxSVG@begingroup@{transform=matrix(0.89442 -0.4472 0.4472 0.89442 -19.93 29.65)} \lxSVG@begingroup@{_scopebegin=1} \lxSVG@begingroup@{stroke-dasharray=none,stroke-dashoffset=0.0pt} \lxSVG@begingroup@{stroke-linecap=round} \lxSVG@begingroup@{stroke-linejoin=round} \lxSVG@drawpath@unclipped{M -2.88 3.32 C -2.35 1.33 -1.18 0.39 0 0 C -1.18 -0.39 -2.35 -1.33 -2.88 -3.32}{fill:none} \lxSVG@closescope \lxSVG@closescope }}{{\lx@inpgf@ignorespaces}}}} \lxSVG@closescope }\lxSVG@closescope }{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}} {{{}}} }{{}{}\lx@inpgf@ignorespaces}{{}{}\lx@inpgf@ignorespaces}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}} {{{}}} } \lx@inpgf@ignorespaces \lxSVG@closescope } \lxSVG@closescope {}{{}}{} {}{}\lxSVG@begingroup@{_scopebegin=1} \color[rgb]{0,0,1}{}\lxSVG@stroke\lxSVG@drawpath@unclipped{M 0 19.69 L 39.37 39.37}{fill:none} \lx@inpgf@ignorespaces{\lxSVG@begingroup@{_scopebegin=1} {}{{}{}\lx@inpgf@ignorespaces}{}{}{}\lx@inpgf@ignorespaces{{}}{{}}{{}{}\lx@inpgf@ignorespaces}{{}{}\lx@inpgf@ignorespaces}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}\lxSVG@begingroup@{_scopebegin=1} {\lxSVG@begingroup@{_scopebegin=1} {{\lx@inpgf@ignorespaces}} {{}{{{\lx@inpgf@ignorespaces}}{{{\lx@inpgf@ignorespaces}}{\lxSVG@begingroup@{_scopebegin=1} \lxSVG@transformcm{0.89442}{0.4472}{-0.4472}{0.89442}{14.04811pt}{21.2503pt}\lxSVG@begingroup@{transform=matrix(0.89442 0.4472 -0.4472 0.89442 19.44 29.4)} \lxSVG@begingroup@{_scopebegin=1} \lxSVG@begingroup@{stroke-dasharray=none,stroke-dashoffset=0.0pt} \lxSVG@begingroup@{stroke-linecap=round} \lxSVG@begingroup@{stroke-linejoin=round} \lxSVG@drawpath@unclipped{M -2.88 3.32 C -2.35 1.33 -1.18 0.39 0 0 C -1.18 -0.39 -2.35 -1.33 -2.88 -3.32}{fill:none} \lxSVG@closescope \lxSVG@closescope }}{{\lx@inpgf@ignorespaces}}}} \lxSVG@closescope }\lxSVG@closescope }{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}} {{{}}} }{{}{}\lx@inpgf@ignorespaces}{{}{}\lx@inpgf@ignorespaces}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}} {{{}}} } \lx@inpgf@ignorespaces \lxSVG@closescope } \lxSVG@closescope {}{{}}{} {}{}\lxSVG@begingroup@{_scopebegin=1} \color[rgb]{0,0,1}{}\lxSVG@stroke\lxSVG@drawpath@unclipped{M 0 -19.69 L 0 19.69}{fill:none} \lx@inpgf@ignorespaces{\lxSVG@begingroup@{_scopebegin=1} {}{{}{}\lx@inpgf@ignorespaces}{}{}{}\lx@inpgf@ignorespaces{{}}{{}}{{}{}\lx@inpgf@ignorespaces}{{}{}\lx@inpgf@ignorespaces}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}\lxSVG@begingroup@{_scopebegin=1} {\lxSVG@begingroup@{_scopebegin=1} {{\lx@inpgf@ignorespaces}} {{}{{{\lx@inpgf@ignorespaces}}{{{\lx@inpgf@ignorespaces}}{\lxSVG@begingroup@{_scopebegin=1} \lxSVG@transformcm{0.0}{1.0}{-1.0}{0.0}{0.0pt}{-0.20001pt}\lxSVG@begingroup@{transform=matrix(0.0 1.0 -1.0 0.0 0 -0.28)} \lxSVG@begingroup@{_scopebegin=1} \lxSVG@begingroup@{stroke-dasharray=none,stroke-dashoffset=0.0pt} \lxSVG@begingroup@{stroke-linecap=round} \lxSVG@begingroup@{stroke-linejoin=round} \lxSVG@drawpath@unclipped{M -2.88 3.32 C -2.35 1.33 -1.18 0.39 0 0 C -1.18 -0.39 -2.35 -1.33 -2.88 -3.32}{fill:none} \lxSVG@closescope \lxSVG@closescope }}{{\lx@inpgf@ignorespaces}}}} \lxSVG@closescope }\lxSVG@closescope }{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}}}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}} {{{}}} }{{}{}\lx@inpgf@ignorespaces}{{}{}\lx@inpgf@ignorespaces}{{{{}{}{{}} }}{{\lx@inpgf@ignorespaces}} {{{}}} } \lx@inpgf@ignorespaces \lxSVG@closescope } \lxSVG@closescope {{}}\lx@inpgf@ignorespaces\hbox{\hbox{{\lxSVG@begingroup@{_scopebegin=1} {{}{}{{ {}{}}}{ {}{}} {{}{{\lx@inpgf@ignorespaces}}}{{}{\lx@inpgf@ignorespaces}}{}{{}{\lx@inpgf@ignorespaces}} {\lx@inpgf@ignorespaces }{{{{\lx@inpgf@ignorespaces}}\lxSVG@begingroup@{_scopebegin=1} \lxSVG@transformcm{1.0}{0.0}{0.0}{1.0}{-4.56248pt}{-31.86942pt}\lxSVG@begingroup@{transform=matrix(1.0 0.0 0.0 1.0 -6.31 -44.1)} \pgfsys@hbox{58}\lxSVG@closescope }}} \lxSVG@closescope }}} {{}}\lx@inpgf@ignorespaces\hbox{\hbox{{\lxSVG@begingroup@{_scopebegin=1} {{}{}{{ {}{}}}{ {}{}} {{}{{\lx@inpgf@ignorespaces}}}{{}{\lx@inpgf@ignorespaces}}{}{{}{\lx@inpgf@ignorespaces}} {\lx@inpgf@ignorespaces }{{{{\lx@inpgf@ignorespaces}}\lxSVG@begingroup@{_scopebegin=1} \lxSVG@transformcm{1.0}{0.0}{0.0}{1.0}{-13.85098pt}{25.73055pt}\lxSVG@begingroup@{transform=matrix(1.0 0.0 0.0 1.0 -19.17 35.6)} \pgfsys@hbox{58}\lxSVG@closescope }}} \lxSVG@closescope }}} {{}}\lx@inpgf@ignorespaces\hbox{\hbox{{\lxSVG@begingroup@{_scopebegin=1} {{}{}{{ {}{}}}{ {}{}} {{}{{\lx@inpgf@ignorespaces}}}{{}{\lx@inpgf@ignorespaces}}{}{{}{\lx@inpgf@ignorespaces}} {\lx@inpgf@ignorespaces }{{{{\lx@inpgf@ignorespaces}}\lxSVG@begingroup@{_scopebegin=1} \lxSVG@transformcm{1.0}{0.0}{0.0}{1.0}{-81.6833pt}{-2.5pt}\lxSVG@begingroup@{transform=matrix(1.0 0.0 0.0 1.0 -113.03 -3.46)} \pgfsys@hbox{58}\lxSVG@closescope }}} \lxSVG@closescope }}} {{}}\lx@inpgf@ignorespaces\hbox{\hbox{{\lxSVG@begingroup@{_scopebegin=1} {{}{}{{ {}{}}}{ {}{}} {{}{{\lx@inpgf@ignorespaces}}}{{}{\lx@inpgf@ignorespaces}}{}{{}{\lx@inpgf@ignorespaces}} {\lx@inpgf@ignorespaces }{{{{\lx@inpgf@ignorespaces}}\lxSVG@begingroup@{_scopebegin=1} \lxSVG@transformcm{1.0}{0.0}{0.0}{1.0}{32.2319pt}{-2.5pt}\lxSVG@begingroup@{transform=matrix(1.0 0.0 0.0 1.0 44.6 -3.46)} \pgfsys@hbox{58}\lxSVG@closescope }}} \lxSVG@closescope }}} \lxSVG@closescope {\lx@inpgf@ignorespaces}{\lx@inpgf@ignorespaces}{\lx@inpgf@ignorespaces}\hss}\lxSVG@discardpath\lxSVG@closescope \hss}}\lxSVG@closescope\endpgfpicture}} For the right pictures, on the one hand, the signs of the crossings of \eta_{1} with the weave edges are -,+,+ ; the h_{-} distance between the bottom and the right decorated flags is (-1)^{\alpha_{i}^{\vee}}h_{3} ; the h_{+} distance between the right and the left decorated flags is s_{i}(h_{3}^{-1})(q-p)^{-\alpha_{i}^{\vee}}h_{1} ; the h_{+} distance between the left and the top decorated flags is h_{1}^{-1}p^{\alpha_{i}^{\vee}}h_{2} . Thus M^{\eta_{1}} is equal to M^{\eta_{1}}=((-1)^{\alpha_{i}^{\vee}}h_{3})^{\mu}\cdot(s_{i}(h_{3}^{-1})(q-p)^{-\alpha_{i}^{\vee}}h_{1})^{s_{i}.\mu}\cdot(s_{i}(h_{1}^{-1})p^{\alpha_{i}^{\vee}}h_{2})^{\mu}=\left(\frac{p}{p-q}\right)h_{2}^{\mu}. On the other hand, the signs of the crossings of \eta_{2} with the weave edges are +,-,+ ; the h_{+} distance between the bottom and the left decorated flags is h_{1} ; the h_{-} distance between the left and the right decorated flags is s(h_{3}^{-1})(q-p)^{-\alpha_{i}^{\vee}}h_{1} ; the h_{+} distance between the right and the top decorated flags is s_{i}(h_{1}^{-1})q^{\alpha_{i}^{\vee}}h_{2} . Thus M^{\eta_{2}} is equal to M^{\eta_{2}}=h_{1}^{\mu}\cdot(s(h_{1}^{-1})(q-p)^{-\alpha_{i}^{\vee}}h_{3})^{s_{i}.\mu}\cdot(s_{i}(h_{1}^{-1})q^{\alpha_{i}^{\vee}}h_{2})^{\mu}=\left(\frac{q}{q-p}\right)h_{2}^{\mu}. Thus M^{\eta_{1}}+M^{\eta_{2}}=\left(\frac{p}{p-q}+\frac{q}{q-p}\right)h_{2}^{\mu}=h_{2}^{\mu}=M^{\eta} . ∎ 4. Applications 4.1. Cluster Theory In the case where a cluster seed can be described by a weave (e.g., Demazure weaves [ CGG+24 ] ), we can use weighted cycles to give a pseudo-topological description of the cluster structure. Definition 4.1 . Let \eta be the weighted cycle representative of a Y-cycle \gamma . We define the cluster \mathcal{X} -variable X_{\gamma}:=M^{\eta} . Definition 4.2 . Let r be the rank of G , let \tau be the number of trivalent weave vertices, and let \beta be the number of boundary base points. Let m:=\tau+r(\beta-1) . Let \{\gamma_{1},\dots,\gamma_{m}\} be a set of linearly independent Y-cycles. Let \{\eta_{1},\dots,\eta_{m}\} be a set of weighted cycles dual to \{\gamma_{1},\dots,\gamma_{m}\} in the sense that \left\langle\eta_{a},\gamma_{b}\right\rangle=\delta_{ab} . We define the cluster \mathcal{A} -variables A_{1},\dots,A_{m} by setting A_{a}:=M^{\eta_{a}} . Remark 4.3 . We will prove in the next section that in the case of Demazure weaves, Definitions 4.1 and 4.2 give the same cluster \mathcal{X} - and \mathcal{A} -variables as in [ CGG+24 ] . Remark 4.4 . By Theorem 3.12 , the rank of the lattice of weighted cycles is precisely m=\tau+r(\beta-1) ; thus, the linearly independent weighted cycles \eta_{1},\dots,\eta_{m} form a sublattice of finite index. If this sublattice is saturated, then we can express the weighted cycle representatives of Y-cycles as a linear combination of \eta_{1},\dots,\eta_{m} , and thus giving a way to express any cluster \mathcal{X} -variable X_{a}=X_{\gamma_{a}} as a product of the cluster \mathcal{A} -variables A_{1},\dots,A_{m} , which must necessarily be X_{a}=\prod_{b}A_{b}^{\epsilon_{ab}} due to the intersection pairing. This is equivalent to the cluster theoretical p -map [ FG09 ] . Remark 4.5 . The mutations of cluster \mathcal{A} -variables and cluster \mathcal{X} -variables are compatible with the local mutation rule for weighted cycles described in Definition 3.33 . For example, note that in Figure 30 , the intersection number between the weighted cycle and the weave edge in the left picture is 1 , and the intersection numbers in both of the right pictures are 1,-1,1 . This coincides with the mutation formula of cluster \mathcal{A} -variables: A^{\prime}_{c}=\frac{\prod_{\epsilon_{ca}>0}A_{a}^{\epsilon_{ca}}+\prod_{\epsilon_{cb}<0}A_{b}^{-\epsilon_{cb}}}{A_{c}} A similar analysis reveals the compatibility with cluster \mathcal{X} -variables mutations as well. 4.2. Demazure Weaves In [ CGGS24 , CGG+24 ] , a special family of weaves called Demazure weaves were introduced, and each Demazure weave was equipped with a collection of special Y-cycles called Lusztig cycles. Let us first recall their definitions below. Definition 4.6 . A Demazure weave is a weave drawn on the rectangle [0,1]\times[0,1] such that: • All external weave edges are incident to the top or the bottom boundary of the rectangle. • No point along any weave edge admits a horizontal tangent line. • Each trivalent weave vertex is incident to two weave edges above it and one below it. • Each tetravalent weave vertex is incident to two weave edges above it and two below it. • Each hexavalent weave vertex is incident to three weave edges above it and three below it. We put one boundary base point at the lower left-hand corner of the rectangle and another one at the lower right-hand corner of the rectangle. Also we assume without loss of generality that all weave vertices in a Demazure weave are at different heights. Because of this definition, all edges in a Demazure weave can be equipped with a downward pointing orientation, and such a choice of orientations on weave edges is an example of the compatible orientations in the sense of Definition 3.18 . Definition 4.7 . Let \mathfrak{w} be a Demazure weave and let v be a trivalent weave vertex in \mathfrak{w} . The Lusztig cycle associated with v is constructed as follows as we scan \mathfrak{w} from top to bottom. • Any weave edge e that begins above v is assigned with \gamma(e)=0 . • The unique weave edge e that begins at v is assigned with \gamma(e)=1 . • For any trivalent weave vertex w below v , the assignments must satisfy \gamma(c)=\min\{\gamma(a),\gamma(b)\} , where c is the unique weave edge that begins at w . • For any tetravalent weave vertex w , the assignments must satisfy \gamma(a)=\gamma(c) and \gamma(b)=\gamma(d) , where a,b,c,d are weave edges incident to w in a cyclic order, with a,b above w and c,d below w . • For any hexavalent weave vertex w , the assignments must satisfy \gamma(d)=\gamma(a)+\gamma(b)-\min\{\gamma(a),\gamma(c)\} , \gamma(e)=\min\{\gamma(a),\gamma(c)\} , and \gamma(f)=\gamma(b)+\gamma(c)-\min\{\gamma(a),\gamma(c)\} , where a,b,c,d,e,f are weave edges incident to w in a cyclic order, with a,b,c above w and d,e,f below w . Recall that in [ CGG+24 ] , each weave edge e is equipped with a labeling (z_{e},u_{e}) such that the two decorated flags on the left and on the right of e are xN and xB_{i}(z_{e})u_{e}^{\alpha_{i}^{\vee}}N where s_{i} is the color of e . Lemma 4.8 . Suppose \eta is a rightward-oriented weighted chain that only intersects the weave edge e with color s_{i} and labeling (z_{e},u_{e}) and suppose the intersection number is n . Then M^{\eta}=u_{e}^{n} . Proof. Note that the intersection is positive. Suppose the chamber weight associated with the right side of \eta is \mu . Then from the intersection number, we know that \left\langle\mu,\alpha_{i}^{\vee}\right\rangle=n . Note that NB_{i}(z_{e})u_{e}^{\alpha_{i}^{\vee}}N=N\overline{s}_{i}u_{e}^{\alpha_{i}^{\vee}}N . Thus, M^{\eta}=\left(h_{+}(xN,xB_{i}(z_{e})u_{e}^{\alpha_{i}^{\vee}})\right)^{\mu}=u_{e}^{\left\langle\mu,\alpha_{i}^{\vee}\right\rangle}=u_{e}^{n}.\qed Let us also recall from [ CGG+24 , Theorem 5.12] that the cluster variables \{A_{a}\} associated with the Demazure weave \mathfrak{w} are uniquely determined by the condition (4.9) u_{e}=\prod_{a}A_{a}^{\gamma_{a}(e)}, where e is any weave edge. Theorem 4.10 . Let \mathfrak{w} be a Demazure weave and let \{\gamma_{a}\} be the collection of Lusztig cycles on \mathfrak{w} . Then there exist a collection of dual weighted relative cycles \{\eta_{a}\} on \mathfrak{w} such that \left\langle\gamma_{a},\eta_{b}\right\rangle=\delta_{ab} and M^{\eta_{a}}=A_{a} for all a , where A_{a} is the cluster variable associated with the Y-cycle \gamma_{a} . Proof. Let p_{a} be the trivalent weave vertex that defines a Y-cycle \gamma_{a} and suppose the color of the weave edges incident to p_{a} is s_{i} . We define the weighted relative cycle \xi_{a} by drawing a short weighted chain going from left to right across the weave edge directly below p_{a} , with chamber weights changing from s_{i}.\omega_{i} to \omega_{i} ; we then extend this weighted chain horizontally at both ends until they reach the boundary of the disk. Note that by construction, \left\langle\gamma_{a},\xi_{a}\right\rangle=1 . Moreover, if we define aa . Now we can go from top to bottom along the weave and define \eta_{a}:=\xi_{a}-\sum_{ba$}.\end{array}\right. As for the merodromy claim, it follows from Lemma 4.8 that M^{\xi_{a}}=\prod_{e\cap\xi_{a}\neq\emptyset}u_{e}^{\left\langle e,\xi_{a}\right\rangle}=A_{a}\prod_{e\cap\xi_{a}\neq\emptyset}\prod_{b