Strong general bound for non-representable matroids

Establish that for every finite set V of dimension vectors, each having at most t coordinates, with k equal to the maximum coordinate appearing in any vector in V, the covering threshold for simple matroids satisfies C_M(V) ≤ binom(t+k+1,k+1).

Background

The paper defines C_M(V) as the least integer N such that a simple matroid on N elements whose restrictions are all covered by V must itself be covered by V. The authors prove the weaker bound C_M(V) ≤ Σ_{i=0}{k+1} ti for arbitrary matroids, while their algebraic argument gives the sharper binomial bound for representable matroids.

The conjecture asks whether the sharper representable-matroid bound remains valid for all matroids, including non-representable ones. The authors note that any extremal counterexample or equality case would likely need substantial structure, because matroids with many small circuits tend to be representable.

References

We conjecture that the representable bound holds in general.

Covering points with planes  (2502.08945 - Dao et al., 13 Feb 2025) in Section 4, “Combinatorial bounds,” immediately after Theorem 4.1 (the theorem labeled “thm:general weak bound”)