Optimal subgroups and applications to nilpotent elements (Q1024622)

Let \(G\) be a connected reductive linear algebraic group over an algebraically closed field of characteristic \(p \geq 0\) acting on the affine algebraic variety \(X\). Let \(x \in X\) have a non-closed orbit, and let \(S\) be a subvariety of which does not contain \(x\) but meets the closure of the orbit of \(X\). Suppose a length function has been chosen on the set \(Y(G)\) of cocharacters of \(G\). There is a morphism \(M(\lambda): \mathbb A^1: \to X\) for every cocharacter \(\lambda\) for which \(\lim_{t\to 0}\lambda(t)\cdot x\) exists; \(\alpha_{S,x}(\lambda)\) is defined to be the degree of the divisor \(M(\lambda)^{-1}(S)\). Kempf has proven that the set \(\Omega_G(x,S)\) of indivisible cocharacters \(\lambda\) for which the ratio of \(\alpha_{S,x}(\lambda)\) to the length of \(\lambda\) is a maximum is non-empty, all the cocharacters in it correspond to the same parabolic subgroup, the unipotent radical of the subgroup acts simply transitively on the set, and that any maximal torus of the subgroup contains a unique element of the set. Now let \(H\) be a reductive subgroup of \(G\) and suppose that the \(H\) orbit of \(x\) is not closed and that \(H\) meets its closure. If \(\Omega_H(x,S)=\Omega_G(x,S) \cap Y(H)\) then \(H\) is called optimal. The author proves that if \(K\) is a \(G\) completely reducible subgroup of \(G\) (i.e., if whenever \(K\) is contained in a parabolic subgroup \(P\) of \(G\) then \(K\) is contained in a Levi subgroup of \(P\)), then if \(x\) above is fixed by \(K\) the subgroup \(H=C_G(K)^0\) is optimal. The author applies this to the case that \(X\) is the Lie algebra \(\mathfrak g\) of \(G\), \(x=e\) is a nilpotent element of \(\mathfrak g\), and \(S=\{0\}\), in which case \(\Omega_G(x,S)\) is denoted simply \(\Omega_G(e)\). A cocharacter \(\lambda\) of \(G\) is said to be associated with \(e\) if \(\text{Ad}(\lambda(t))x=t^2x\) for all \(t \in k^*\) and there is a Levi subgroup \(L\) of some parabolic subgroup of \(G\) such that the image of \(\lambda\) is contained in the derived group of \(L\), such that \(e\) belongs to the Lie algebra of \(L\), and every torus in \(C_L(e)\) is central in \(L\). The set of cocharacters associated with \(e\) is denoted \(\Omega_G^a(e)\). The author proves that if \(H\) is a reductive subgroup of \(G\) with Lie algebra \(\mathfrak h\) and \(e \in \mathfrak h\) is nilpotent, then \(\Omega_H^a(e)=\Omega_G^a(e) \cap Y(H)\) if and only if \(\Omega_H(e)=\Omega_G(e) \cap Y(H)\), provided \(p\) is good for \(G\) and \(H\). Combining this result with his result on optimality, this shows that \(\Omega_H^a(e)=\Omega_G^a(e) \cap Y(H)\) whenever \(H=C_G(K)^0\) for \(K\) a reductive subgroup of \(G\).