Two results on centralisers of nilpotent elements (Q2464516)

In the paper [J. Pure Appl. Algebra 206, 123--140 (2006; Zbl 1106.20036)], \textit{G. McNinch} proved the following result: Let \(X\) and \(Y\) be two commuting nilpotent elements of a semisimple Lie algebra \(\mathfrak g=\text{Lie}\;G\) over an algebraically closed field \(k\) of characteristic zero. Then for all but finitely many points \((a:b)\in\mathbb P^1_k\), both \(X\) and \(Y\) belong to the Lie algebra \((\mathfrak g_{aX+bY})^u\) of the uniponent radical of the centralizer of \(aX+bY\) in \(\mathfrak g\). In the first part of this note, the author gives a short alternative proof of this statement, which uses a result of \textit{A. Premet} from [Invent. Math. 154, 653--683 (2003; Zbl 1068.17006)]. The second part of this paper is devoted to self-large nilpotent elements in \(\mathfrak g\). By definition, a nilpotent element is self-large if its \(G\)-orbit \(G\cdot e\) is the largest nilpotent orbit meeting its stabilizer \(\mathfrak g_e\). The characterization of such elements provided by the author is as follows. Let \(\{e,f,h\}\) be an \(\mathfrak{sl}_2\)-triple containing \(e\), and let \(\mathfrak g_e=\bigoplus_{i\geq 0}\mathfrak g_e(i)\) be the grading determined by \(h\). Then \(e\) is self-large iff \(\mathfrak g_e(0)\) is the Lie algebra of a torus and \(\mathfrak g_e(1)=0\).