Hessenberg varieties and Poisson slices

Abstract

This work pursues a circle of Lie-theoretic ideas involving Hessenberg varieties, Poisson geometry, and wonderful compactifications. In more detail, one may associate a symplectic Hamiltonian $G$-variety $\mu:G\times\mathcal{S}\longrightarrow\mathfrak{g}$ to each complex semisimple Lie algebra $\mathfrak{g}$ with adjoint group $G$ and fixed Kostant section $\mathcal{S}\subseteq\mathfrak{g}$. This variety is one of Bielawski's hyperk\"ahler slices, and it is central to Moore and Tachikawa's work on topological quantum field theories. It also bears a close relation to two log symplectic Hamiltonian $G$-varieties $\overline{\mu}_{\mathcal{S}}:\overline{G\times\mathcal{S}}\longrightarrow\mathfrak{g}$ and $\nu:\mathrm{Hess}\longrightarrow\mathfrak{g}$. The former is a Poisson transversal in the log cotangent bundle of the wonderful compactification $\overline{G}$, while the latter is the standard family of Hessenberg varieties. Each of $\overline{\mu}$ and $\nu$ is known to be a fibrewise compactification of $\mu$. We exploit the theory of Poisson slices to relate the fibrewise compactifications mentioned above. Our main result is a canonical $G$-equivariant bimeromorphism $\mathrm{Hess}\cong\overline{G\times\mathcal{S}}$ of varieties over $\mathfrak{g}$. This bimeromorphism is shown to be a Hamiltonian $G$-variety isomorphism in codimension one, and to be compatible with a Poisson isomorphism obtained by B\u{a}libanu. We also show our bimeromorphism to be a biholomorphism if $\mathfrak{g}=\mathfrak{sl}_2$, and we conjecture that this is the case for arbitrary $\mathfrak{g}$. We conclude by discussing the implications of our conjecture for Hessenberg varieties.