Kazhdan–Laumon finite cohomological dimension conjecture

About 3 years old · traced to

Let GG be a semisimple algebraic group split over a finite field, let A\mathcal{A} be the abelian category of glued perverse sheaves in the Kazhdan–Laumon construction, and let A,BA,B be objects of A\mathcal{A}. Kazhdan–Laumon conjecture. The category A\mathcal{A} has finite cohomological dimension: there exists an nn such that

Ext⁡i(A,B)=0\operatorname{Ext}^i(A,B)=0

whenever i>ni>n. This conjecture was disproved by Bezrukavnikov and Polishchuk in the case G=SL⁡3G=\operatorname{SL}_3, so it is no longer open.

References

Primary source

Calder Morton-Ferguson, “Polishchuk's conjecture and Kazhdan-Laumon representations”, arXiv:2309.13462 (2025).

Progress summary

Refreshed
Claimed solved

The conjecture is false: a published counterexample shows that the relevant category does not have finite cohomological dimension.

Kazhdan and Laumon proposed the conjecture in 1988. It was disproved by Bezrukavnikov and Polishchuk in 2001 for G=SL⁡3G=\operatorname{SL}_3.

Known results

  • The simple object corresponding to the constant sheaf has infinite cohomological dimension (Bezrukavnikov and Polishchuk, 2001).
  • The weaker localized-K0K_0 statement was established in types AnA_n with n≥3n\ge 3 and B2B_2 (Polishchuk, 2001).

Later replacement result, 2023

A 2023 paper proved the localized-K0K_0 replacement in all types and established well-definedness of the Kazhdan–Laumon construction by this alternative route; it does not restore the disproved finite-cohomological-dimension conjecture.

Current status (as of September 2026): The original conjecture is settled negatively by the SL⁡3\operatorname{SL}_3 counterexample; the replacement conjecture and well-definedness result are established more broadly.

Sources

Solutions 0

No solutions have been posted yet.