An algorithm for Aubert-Zelevinsky duality à la Mœglin-Waldspurger

Let $F$ be a locally compact non-Archimedean field of characteristic $0$, and let $G$ be either the split special orthogonal group ${\mathrm{S}\mathrm{O}}_{2n+1}(F)$ or the symplectic group ${\mathrm{S}\mathrm{p}}_{2n}(F)$. The goal of this paper is to give an explicit description of the Aubert-Zelevinsky duality for $G$ in terms of Langlands parameters. We present a new algorithm, inspired by the Moeglin-Waldspurger algorithm for ${\mathrm{G}\mathrm{L}}_n(F)$, which computes the dual Langlands data in a recursive and combinatorial way. Our method is simple enough to be carried out by hand and provides a practical tool for explicit computations. Interestingly, the algorithm was discovered with the help of machine learning tools, guiding us toward patterns that led to its formulation.