On the Kostant conjecture for Clifford algebra

TL;DR

This paper proves the equivalence of two filtrations on a Cartan subalgebra related to the Kostant conjecture, connecting Clifford algebra methods with classical Lie algebra projections.

Contribution

It demonstrates that Joseph's recent result on filtrations is equivalent to Kostant's conjecture and relates different projection-based filtrations in Lie algebra theory.

Findings

Joseph's filtration coincides with the Kostant filtration

Standard Harish-Chandra projection induces the same filtration

Proves equivalence of different approaches to Kostant's conjecture

Abstract

Let g be a complex simple Lie algebra, and h be a Cartan subalgebra. In the end of 1990s, B. Kostant defined two filtrations on h, one using the Clifford algebras and the odd analogue of the Harish-Chandra projection $hc: Cl(g) \to Cl(h)$, and the other one using the canonical isomorphism $\check{h} = h^*$ (here $\check{h}$ is the Cartan subalgebra in the simple Lie algebra corresponding to the dual root system) and the adjoint action of the principal sl2-triple. Kostant conjectured that the two filtrations coincide. The two filtrations arise in very different contexts, and comparing them proved to be a difficult task. Y. Bazlov settled the conjecture for g of type A using explicit expressions for primitive invariants in the exterior algebra of g. Up to now this approach did not lead to a proof for all simple Lie algebras. Recently, A. Joseph proved that the second Kostant filtration coincides with the filtration on h induced by the generalized Harish-Chandra projection $(Ug \otimes g)^g \to Sh \otimes h$ and the evaluation at $\rho \in h^*$. In this note, we prove that Joseph's result is equivalent to the Kostant Conjecture. We also show that the standard Harish-Chandra projection $Ug \to Sh$ composed with evaluation at $\rho$ induces the same filtration on h.