Classical theta lifts for higher metaplectic covering groups (Q2216470): Difference between revisions

The aim of the paper under review is to explore possible theta liftings for coverings of higher degree. Compared with the well-known twofold case, the obvious hurdle here is the lack of Weil representation for higher covering, which plays the role of integral kernel. In this paper, one considers an odd number \(r \geq 1\), a number field \(F\) with all \(2r\)-th roots of unity, and fixes an additive character \(\psi\) for \(F \backslash \mathbb{A}_F\). Fix \(k, n\) and take \((G_1, G_2) = (\mathrm{SO}_k, \mathrm{Sp}_{2n})\), together with embeddings \[ \mathrm{Sp}_{2n + k(r-1)} = \mathrm{Sp}(W_2) \stackrel{\iota_2}{\hookleftarrow} G_1 \times G_2 \stackrel{\iota_1}{\hookrightarrow} \mathrm{Sp}(W_1) = \mathrm{Sp}_{2nk}, \] where \(\iota_1\) is the usual \(\otimes\)-embedding and \[ \iota_2(h, g) = \mathrm{diag}(h, \ldots, g, h^*, \ldots, h^*) \] with \(h\) appeared \(\frac{r-1}{2}\) times and \(h^*\) is defined to make the right-hand side symplectic. Moreover, there is an explicitly given unipotent subgroup \(U \subset \mathrm{Sp}(W_2)\), normalized by the image of \(\iota_2\), together with an algebraic homomorphism \(l: U \to \mathcal{H}(W_1)\) into the Heisenberg group. There is also a character \(\psi_U\) for \([U]\). One can define homomorphisms \[ \iota_2^{(r)}: \widetilde{\mathrm{SO}}_k^{(r)} \times \widetilde{\mathrm{Sp}}_{2n}^{(r)} \to \widetilde{\mathrm{Sp}}_{2n + k(r-1)}^{(r)} \] and \[ \iota_1^{(2)}: \mathrm{SO}_k \times \widetilde{\mathrm{Sp}}^{\kappa}_{2n} \to \widetilde{\mathrm{Sp}}^{(2)}_{2nk} \] where the \(\sim\) and exponents \((r)\) denotes appropriate \(r\)-fold Brylinski-Deligne coverings, and \(\kappa \in \{1, 2\}\) depends on the parity of \(k\) as in the classical theta correspondence. From all these data, one can use Weil's representation \(\omega_\psi\) on \(\widetilde{\mathrm{Sp}}^{(2)}_{2nk}\) and the genuine theta representation \(\Theta^{(r)}_{2n+k(r-1)}\) on \(\widetilde{\mathrm{Sp}}^{(r)}_{2n + k(r-1)}\) to construct an integral kernel, which also involves taking \((U, \psi_U)\)-Fourier coefficients, and then define the theta lifting from \(\widetilde{\mathrm{Sp}}^{(r\kappa)}_{2n}(\mathbb{A}_F)\) to \(\widetilde{\mathrm{SO}}^{(r)}_k(\mathbb{A}_F)\). These liftings fit into a tower. Under the Conjecture 1 on the maximal unipotent orbits supporting Fourier coefficients of \(\Theta^{(r)}_{2n+k(r-1)}\), which is verified for \(r \in \{3, 5\}\), Theorem 2 shows the cuspidality of the first occurrence. Over local fields, these constructions admit local versions defined using \(\mathrm{Hom}\)-spaces, as usual. As the authors indicated, they can probably be taken farther. More broadly, one may expect to generalize automorphic constructions to Brylinski-Deligne coverings.