Varieties of nilpotent elements for simple Lie algebras. II: Bad primes. (Q2577536)

From the text: Let \(G\) be a simple algebraic group over an algebraically closed field \(k\) of characteristic \(p>0\) and let \(\mathfrak g\) be the (restricted) Lie algebra of \(G\) with \(p\)th power map \([p]\). The maximal ideal spectrum of the cohomology ring of the restricted enveloping algebra Maxspec\((H^*(u(\mathfrak g),k))\) can be identified with the variety \(\mathcal N_1(\mathfrak g)=\{x\in\mathfrak g:x^{[p]}=0\}\). When the characteristic of the field is a good prime, this variety was first described as the closure of a certain Richardson orbit by Carlson, Lin, Nakano, and Parshall. Their methods used the techniques developed by Nakano, Parshall, and Vella which involved the verification of a conjecture of Jantzen on the support varieties of Weyl modules. More recently, the [current] authors investigated a more general question. Let \(\rho\) be a finite-dimensional representation of \(\mathfrak g\) which is realized as the derivative of a representation of \(G\). For \(r\geq 0\), set \(\mathcal N_{r,\rho}(\mathfrak g)=\{x\in\mathcal N(\mathfrak g):\rho(x)^r=0\},\) where \(\mathcal N(\mathfrak g)\) is the variety of nilpotent elements of \(\mathfrak g\). The variety \(\mathcal N_{r,\rho}(\mathfrak g)\) is a \(G\)-invariant subvariety of \(\mathcal N(\mathfrak g)\). Since there are finitely many \(G\)-orbits on \(\mathcal N(\mathfrak g)\), one can ask, how can this variety be expressed as a finite union of orbit closures. The authors determined \(\mathcal N_{r,\rho}(\mathfrak g)\) when \(\rho\) is either a minimal dimensional irreducible representation or the adjoint representation of \(\mathfrak g\) and the characteristic of the field \(k\) is a good prime relative to the underlying root system of \(\mathfrak g\). One of the main objectives of this paper is to determine the restricted nullcone \(\mathcal N_1(\mathfrak g)\) over the fields of bad characteristic. Since \(\mathcal N_1(\mathfrak g)\cong\) Maxspec\((H^*(u(\mathfrak g),k))\), the determination of \(\mathcal N_1(\mathfrak g)\) for bad primes is a key step in the study of support varieties for restricted Lie algebras. We have recently used our determination of \(\mathcal N_1(\mathfrak g)\) to calculate support varieties of Weyl modules over fields of bad characteristic. Let \(\rho\) be a finite-dimensional representation of \(G\) and let \(\mathcal U(G)\) be the variety of unipotent elements in \(G\). One can define a subvariety \(\mathcal U_{r,\rho}(G)\) of \(\mathcal U(G)\), which is analogous to \(\mathcal N_{r,\rho}(\mathfrak g)\), by letting \(\mathcal U_{r,\rho}(G)=\{x\in\mathcal U(G): (\rho(x)-1)^r=0\}.\) An analog of the restricted nullcone for the unipotent variety would be \(\mathcal U_1(G)=\{x\in\mathcal U(G): x^p=1\}\). We will refer to this as the restricted unipotent variety. Since there are finitely many unipotent classes it is also reasonable to try and describe \(\mathcal U_{r,\rho}(G)\) as a finite union of closures of such classes. From our computations of \(\mathcal N_{r,\rho}(\mathfrak g)\) and \(\mathcal U_{r,\rho}(G)\), we set \(r=\rho\) to obtain concrete descriptions of the restricted nullcone \(\mathcal N_1(\mathfrak g)\) and the restricted unipotent variety \(\mathcal U_1(G)\) for bad primes. Our results in conjunction with work [by Carlson, Lin, Nakano, and Parshall] demonstrate that these varieties are indeed irreducible for all primes, thus answering an old question posed by Friedlander and Parshall. Furthermore, in this process, we establish a remarkable fact: there is an order preserving bijection between the nilpotent orbits in \(\mathcal N_1(\mathfrak g)\) and unipotent classes in \(\mathcal U_1(G)\) for all primes.'' Part I, cf. J. Algebra 280, No. 2, 719--737 (2004; Zbl 1063.17007).