Abelian ideals of a Borel subalgebra and subsets of the Dynkin diagram (Q765708)

In 1965, B. Kostant began the study of all commutative subalgebras of a complex simple Lie algebra \(\mathfrak{g}\). A central object in his research was the family \(\mathfrak{Ab}(\mathfrak{g})\) of all abelian ideals in a fixed Borel subalgebra \(\mathfrak{b}\) of \(\mathfrak{g}\). More than thirty years later, he wrote a summary of those results [see \textit{B. Kostant}, Int. Math. Res. Not. 1998, No. 5, 225--252 (1998; Zbl 0896.17002)], including a result of D. Peterson stating that the number of abelian ideals of \(\mathfrak{b}\) is \(2^l\), for \(l\) the rank of the algebra. In the paper under review, the author observes empirically a refinement of such result: the number of abelian ideals with \(k\) generators equals the number of subsets of the Dynkin diagram with \(k\) connected components. (Any abelian ideal \(\mathfrak{a}\) is identified with the set of positive roots such that \(\mathfrak{a}\) is the sum of the corresponding root spaces, and one of such roots is said to be a \textit{generator} of \(\mathfrak{a}\) if it is minimal with the usual root order relative to a set of simple roots.) The objective of this work is to understand such coincidence by finding a bijection between those two sets. The aim is achieved when \(\mathfrak{g}\in\{\mathfrak{sl}(n+1),\mathfrak{sp}(2n)\}\). Another bijection between \(\mathfrak{Ab}(\mathfrak{g})\) and the subsets of the Dynkin diagram is also constructed for \(\mathfrak{g}\) arbitrary (simple), but, unfortunately, it does not respect the number of generators and connected components. More precisely, it is a bijection among the subsets of the Dynkin diagram and the so called \textit{minuscule} elements of the affine Weyl group, thanks to a connection of such elements with the abelian ideals proved by Kostant-Peterson. A new proof of this relation is offered to the reader in an appendix.