The topology of moduli spaces of group representations: the case of compact surface (Q545199): Difference between revisions

Let G be a connected reductive affine algebraic group over the field of complex numbers, and K be its maximal compact subgroup. It was proved by \textit{C. Florentino} and \textit{S. Lawton} [Math. Ann. 345, No. 2, 453--489 (2009; Zbl 1200.14093)] that \(\mathrm{Hom}(F,K)/K\) is a strong deformation retraction of \(\mathrm{Hom}(F,G)//G\), where \(F\) is a free group. For a Riemann surface \(X\) whose fundamental group is isomorphic to a free group, this result can be rephrased that the moduli space of flat \(K\)-connections on \(X\) is a strong deformation retraction of the moduli space of flat \(G\)-connections on \(X\). The purpose of the paper under review is to prove that it is not the case when \(X\) is compact: i.e., if \(X\) is a compact Riemann surface of genus at least two and \(G\) is nontrivial semisimple, they prove that \(\mathrm{Hom}(\pi_{1}(X),K)/K\) is NOT a deformation retraction of \(\mathrm{Hom}(\pi_{1}(X),G)//G\); equivalently, the moduli space \(M_{G}(X)\) of topologically trivial semistable principal \(G\)-bundles on \(X\) is not a deformation retraction of the moduli space \(H_{G}(X)\) of Higgs \(G\)-bundles \((E_{G},\theta)\) on \(X\), where \(E_G\) is a principal \(G\)-bundle on \(X\) and \(\theta\in\mathrm{H}^0(X,\mathrm{ad}(E_G)\otimes K_X)\) is a Higgs field on \(E_G\). Actually it is proven that the singular cohomology groups of these two moduli spaces are not isomorphic. In order to prove it, they explicitly construct a deformation retraction of \(H_{G}(X)\) to a proper algebraic subset \(\mathcal{N}\), the nilpotent cone, which is defined as the null-set of the \textit{N. Hitchin} map \(\mathcal{H}:H_{G}(X)\to\bigoplus_{j=1}^{\ell}\mathrm{H}^0(X,K_X^{\otimes n_j})\) [Duke Math. J. 54, 91--114 (1987; Zbl 0627.14024)] (See Theorem 2.1). Explicit description and properness of the Hitchin map are used in the construction. Since \(M_{G}(X)\) is known to be an irreducible projective variety of dimension \(\dim{G}\cdot (g-1)\), \(\mathrm{H}^{\dim{G}\cdot (g-1)}(M_{G}(X),\mathbb{Z})\) equals to \(\mathbb{Z}\). On the other hand, \(\mathcal{N}\) is known to be equidimensional of dimension \(\dim{G}\cdot (g-1)\) and its irreducible components are parametrized by conjugacy classes of nilpotent elements in \(\mathrm{Lie}(G)\) [\textit{G. Laumon}, Duke Math. J. 57, No.2, 647-671 (1988; Zbl 0688.14023)]. There are at least two such classes, so \(\mathrm{H}^{\dim{G}\cdot (g-1)}(\mathcal{N},{Z})\), which is isomorphic to \(\mathrm{H}^{\dim{G}\cdot (g-1)}(M_{G}(X),{Z})\), is of rank at least two. This concludes the proof.