Atkinson-Lloyd conjecture for primitively intransitive spaces of linear operators

Let UU and VV be finite-dimensional vector spaces over a field F\mathbb{F}. Let S\mathcal{S} be a linear subspace of Hom(U,V)\mathrm{Hom}(U,V). We say that it is intransitive when Sx≠V\mathcal{S} x \neq V for all x∈Ux \in U, and we say that it is primitively intransitive when, in addition, there is no proper linear subspace WW of VV such that πV/WS\pi^{V/W} \mathcal{S} is an intransitive subspace of Hom(U,V/W)\mathrm{Hom}(U,V/W), where πV/W:V↠V/W\pi^{V/W} : V \twoheadrightarrow V/W stands for the canonical projection.

Atkinson and Lloyd have proved that if ∣F∣≥n|\mathbb{F}| \geq n and S\mathcal{S} is primitively intransitive then dim⁡S≤n(n−1)2\dim \mathcal{S} \leq \frac{n(n-1)}{2}. The conjecture states that the result holds without the cardinality assumption on F\mathbb{F}.

References

Progress summary

Refreshed
Claimed progress

The original bound is known for sufficiently large fields, while a recent paper appears to extend related results but does not clearly settle the conjecture for every field.

Atkinson and Lloyd conjectured that every primitively intransitive operator space satisfies the stated dimension bound without any restriction on the field. Their published theorem imposed the hypothesis ∣F∣≥n|\mathbb{F}|\geq n.

Known results

  • Atkinson and Lloyd: if ∣F∣≥n|\mathbb{F}|\geq n, then dim⁡S≤n(n−1)/2\dim\mathcal{S}\leq n(n-1)/2.

Recent related result, reported September 2026

The paper Spaces of matrices with few eigenvalues (II) states an Atkinson-type theorem for primitively intransitive spaces and says its main results hold beyond the usual field-cardinality assumptions, apart from the cases ∣F∣≤2|\mathbb{F}|\leq 2. The retrieved text does not explicitly identify this as a proof of the exact Atkinson–Lloyd conjecture, so this is claimed progress rather than a verified resolution.

Current status (as of September 2026): The conjecture is proved under ∣F∣≥n|\mathbb{F}|\geq n, while its unrestricted formulation remains open; a recent paper may cover related cases for ∣F∣>2|\mathbb{F}|>2 but does not explicitly establish this exact statement.

Sources

Solutions 0

No solutions have been posted yet.