← All problems

Frenkel’s root-multiplicity bound conjecture

Let g\mathfrak{g} be the rank-three hyperbolic Kac–Moody Lie algebra of type A1++A_1^{++}, with invariant bilinear form (⋅,⋅)(\cdot,\cdot), and let p(n)p(n) denote the ordinary partition function. The conjecture asserts that for every root α\alpha of g\mathfrak{g}, dim⁡gα≤p ⁣(1−(α,α)2)\dim \mathfrak{g}_\alpha \le p\!\left(1-\frac{(\alpha,\alpha)}{2}\right).

References

Progress summary

Refreshed
Open · solution reported

A September 2026 preprint claims to prove the conjecture for the three-dimensional example, but the claim has not been independently checked.

Frenkel’s conjecture bounds every root-space multiplicity in the rank-three hyperbolic algebra g=A1++\mathfrak{g}=A_1^{++} by a partition number. It was stated explicitly in 1982, with proceedings published in 1985.

Known results

  • Feingold and Frenkel obtained exact level-one and level-two information (1983).
  • Kang derived homological multiplicity formulas through level five (1993–1994).
  • The general conjecture is known in dimension 2626 but fails for E10E_{10} (Kac, Moody, and Wakimoto, 1988).
  • The rank-three case was still recorded as open in 2015.

September 28, 2026 claimed proof

Gabriel B. Legros’s preprint claims a computer-assisted proof for every root of A1++A_1^{++}, combining analytic estimates, Weyl reduction, and finite computation. The preprint also says that GPT6 Astra substantially assisted development of critical ingredients. No independent expert assessment or referee verification was found, so this remains an unverified claim.

Current status (as of October 2026): The bound for A1++A_1^{++} is claimed proved by a September 2026 preprint, but remains unverified; earlier general formulations include a known E10E_{10} counterexample.

Sources

Solutions 0

No solutions have been posted yet.