Generalized linear-period classification conjecture and Lapid–Prasad conjecture

Let D\mathrm{D} be a quaternion division algebra over a non-Archimedean local field F\mathrm{F} of characteristic zero, let Gn=GLn(D)\mathrm{G}_n=\mathrm{GL}_n(\mathrm{D}), and let H1,n−1≃G1×Gn−1\mathrm{H}_{1,n-1}\simeq \mathrm{G}_1\times \mathrm{G}_{n-1} be the block-diagonal subgroup of Gn\mathrm{G}_n. For s∈Rs\in\mathbb{R}, define χs(diag⁡(g1,g2))=ν(g1)2sν(g2)−2s\chi_s(\operatorname{diag}(g_1,g_2))=\nu(g_1)^{2s}\nu(g_2)^{-2s}. The conjecture is that, for every n>2n>2, an irreducible smooth representation π\pi of Gn\mathrm{G}_n satisfies Hom⁡H1,n−1(π,χs)≠0\operatorname{Hom}_{\mathrm{H}_{1,n-1}}(\pi,\chi_s)\neq 0 if and only if its Langlands parameter L(π)\mathfrak{L}(\pi) contains a Weil--Deligne subrepresentation isomorphic to L(ν−2s)\mathfrak{L}(\nu^{-2s}), of dimension 2n−42n-4, such that the resulting four-dimensional quotient is the Langlands parameter of either the trivial representation of G2\mathrm{G}_2, in which case s=±n−22s=\pm\frac{n-2}{2}, or an irreducible infinite-dimensional representation of G2\mathrm{G}_2 admitting an H1,1\mathrm{H}_{1,1}-invariant functional.

References

Primary source

arXiv

Additional references

Progress summary

Refreshed
Claimed progress

A recent preprint reports the classification in two small cases, but the general conjecture remains unproved.

The conjecture predicts which representations admit generalized linear periods, with a formulation tied to Langlands parameters and the Lapid–Prasad framework. Its all-nn classification is explicitly presented as conjectural.

September 2026 low-rank result

On September 28, 2026, a report citing work of Prem Dagar and Hariom Sharma stated that the classification is proved for n=3n=3 and n=4n=4 for GL⁡n(D)\operatorname{GL}_n(D), while the all-nn statement remains conjectural. The related framework reduces broader linear-period questions to an essentially square-integrable case but does not establish this classification.

Current status (as of September 2026): The cases n=3n=3 and n=4n=4 are reported as proved, but the classification for arbitrary nn remains open and the reported low-rank advance is unverified here.

Sources

Solutions 0

No solutions have been posted yet.