Topological constructions of tensor fields on moduli spaces (Q2237390)

Let \(R\) be a commutative ring. This paper constructs a quasi-Lie algebra structure for a certain \(R\)-algebra \(A\), arising from the fundamental group of a quasi-surface. This is a generalization of Goldman's construction of Lie algebras (the Goldman algebra) arising from surfaces [\textit{W. M. Goldman}, Invent. Math. 85, 263--302 (1986; Zbl 0619.58021)]. Let \(X\) be a compact orientable surface (with possible boundary \(\partial X\)) and \(\pi\) its fundamental group. Let \(\hat{\pi}\) be the set of conjugacy classes (homotopy classes of loops) of \(\pi\) and \(A = \mathbb{Z} \hat{\pi}\) be the free abelian group generated by \(\hat{\pi}\). By considering conjugacy classes, the base point of \(\pi\) becomes irrelevant for the following definition: Given two classes \(a, b \in \hat{\pi}\), there is a well-defined bracket \[ [a,b] = \sum_{p \in a \cap b} \epsilon(p; a, b) a_p b_p, \] where all intersections in \(a \cap b\) are transverse, \(\epsilon(p; a, b)\) is the oriented intersection number of \(a\) and \(b\) at \(p\) and \(a_p b_p\) is the loop starting at \(p\) going around \(a\), followed by going around \(b\). Then this bracket defines a Lie algebra structure on \(A\). The Lie algebra \(A\) is motivated by character varieties [loc. cit.]. Let \(G\) be a linear algebraic group over \(\mathbb{C}\). The group \(G\) acts on the representation variety \(\Hom(\pi, G)\) by equivalence of representations and the \(G\)-character variety is the categorical quotient \(\mathcal{M} := \Hom(\pi,G)/G\), which may be singular. Let \([\rho] \in \mathcal{M}\) and \(a \in \hat{\pi}\). Then \(f_a([\rho]) = \mathrm{trace}(\rho(a))\) is a function in \(C^\infty(\mathcal{M})\). The linear function \[ A \to C^\infty(\mathcal{M}), \qquad a \mapsto f_a \] makes (the smooth part of) \(\mathcal{M}\) a Poisson manifold. The symplectic leaves are the character varieties with fixed monodromy at the boundary circles. The paper under review generalizes the construction of \(A\) for what the author calls quasi-surfaces \(X\) (A typical example is to glue a finite subset of the boundary of a surface). To do this, the author introduces the concept of a quasi-Lie algebra: \((L, [-, -], [-, -, -])\) is a quasi-Lie algebra if \((L, [-, -])\) is a Lie algebra, the triple bracket \([-, -, -]\) is cyclic symmetric and the Jacobi identity is replaced by the quasi-Jacobi identity: \[ [[x,y],z] + [[y, z],x] + [[z,x],y] = [x,y,z]-[y,x,z]. \] The main theorem is as follows: Suppose \(X\) is a quasi-surface with fundamental group \(\pi\). Then \(A := R\hat{\pi}\) has a quasi-Lie algebra structure. The triple bracket arises from the existence of certain objects on \(X\) that the author identifies as gates and that \((A, [-, -])\) is a Lie algebra if every loop in \(\hat{\pi}\) may be deformed away from these gates.