Approximate counting of common matroid bases

A matroid abstracts which subsets of a finite ground set are independent. Given independence tests for two matroids, can the number of subsets that are bases of both be approximated efficiently?

Why it matters and what counts as progress

Why it matters: The ability to recognize a base or find one common base does not directly give the number of common bases. The oracle model asks what can be done without a matrix representation.

What counts as progress: Give a relative-error estimate and failure bound with polynomial work and oracle calls, including the empty-intersection case. State the input oracle, rank and running-time assumptions.

How the problem developed

  1. 6 Oct 2026 · Reported common-base FPRAS [1]

    OpenAI’s internal model Two finite equal-rank matroids supplied by independence oracles admit an FPRAS for their number of common bases. The algorithm returns zero when there are none and has polynomial oracle and bit bounds on every run in the encoded input, ε−1\varepsilon ^{-1} and log⁡δ−1\log \delta ^{-1}.

AI contribution

Approximate counting of common bases of two matroids

OpenAI

Two finite equal-rank matroids supplied by independence oracles admit an FPRAS for their number of common bases. The algorithm returns zero when there are none and has polynomial oracle and bit bounds on every run in the encoded input, ε−1\varepsilon ^{-1} and log⁡δ−1\log \delta ^{-1}.

AI involvementAI-led
ImportanceNotable
Technical verificationFormal (Lean)
Human understandingFew
Paper date:

Sources

  1. Approximate counting of common bases of two matroids
  2. OpenAI mathematics release
  3. OpenAI’s Lean scope for this manuscript
Publication history