Poisson geometry of the moduli of local systems on smooth varieties (Q824252): Difference between revisions

The authors examine the Poisson geometry of moduli spaces of local systems. Let \(X\) be a smooth complex variety of dimension \(d\) and \(G\) a reductive group. If \(X\) is curve than it is well-known that the moduli space of \(G\)-local systems on \(X\) carries a canonical Poisson structure, whose symplectic leaves are moduli of \(G\)-local systems whose monodromy at infinity is fixed (up to conjugacy). The authors give a natural extension of this result to the case of higher dimensions, however even the correct formulation now invariably involves derived geometry. If \(d > 1\) then the moduli space of \(G\)-local systems is naturally replaced by a derived moduli stack \(\mathrm{Loc}_G(X)\) of \(G\)-local systems (note that this takes into account the whole homotopy type of \(X\) as opposed to just the fundamental group). It follows from earlier results of the authors and their collaborators that for any compact oriented (real) manifold \(M\) there is a \((2 - \dim_{ \mathbb R} M)\)-shifted symplectic structure on \(\mathrm{Loc}_G(M)\) and thus a non-degenerate \((2 - \dim_{ \mathbb R} M)\)-shifted Poisson structure. In this new paper the authors show that if \(X\) is a smooth complex variety, not necessarily proper, \(\mathrm{Loc}_G(X)\) carries a canonical \((2-2d)\)-shifted Poission structure. They moreover describe some generalized symplectic leaves of the foliation if \(X\) admits a smooth compactification whose divisor at infinity is simple normal crossing with at most double intersections. Analogously to the case of curves these leaves are given by derived moduli of \(G\)-local system with fixed local monodromy at infinity as long as a technical condition the authors call `strictness' is satisfied. As the authors say this is a first step towards understanding moduli of local systems on higher dimensional open varieties, with nonproper generalizations of Simpson's nonabelian Hodge theory as a key long term motivation. The key ingredient in the proof is the restriction map to the boundary at infinity. Any smooth complex algebraic variety has well-defined boundary at infinity \(\partial X\) that is a compact manifold of real dimension \(2d-1\). Now \(\mathrm{Loc}_G(\partial G)\) has a shifted symplectic structure and by results of Calaque the restriction map is Lagrangian [\textit{D. Calaque}, Contemp. Math. 643, 1--23 (2015; Zbl 1349.14005)]. Then this induces the Poisson structure on \(\mathrm{Loc}_G(X)\) by \textit{V. Melani} and \textit{P. Safronov} [Sel. Math., New Ser. 24, No. 4, 3061--3118 (2018; Zbl 1461.14006); Sel. Math., New Ser. 24, No. 4, 3119--3173 (2018; Zbl 1440.14004)]. The characterization of symplectic leaves is more involved, already the notion of `fixing the monodromy at infinity' is more complicated than in the 1-dimensional case. The paper includes useful discussions of algebraic descriptions of the moduli of \(G\)-local systems and of the boundary at infinity of a smooth complex variety. The special case that \(X\) is a curve is discussed in detail. The authors expect (but do not show) that in this case the 0-shifted symplectic structure they construct agrees with the one that is known from the literature.