← All problems

Frankl’s Intersection–Union Product Conjecture

Let m(n,p,q)m(n,p,q) be the maximum cardinality of a family F⊆2[n]\mathcal{F}\subseteq 2^{[n]} such that, for every distinct A,B∈FA,B\in\mathcal{F}, ∣A∩B∣≥p|A\cap B|\ge p and ∣[n]∖(A∪B)∣≥q|[n]\setminus(A\cup B)|\ge q. For all integers n,p,qn,p,q with 1≤p,q1\le p,q and p+q≤np+q\le n, the conjecture asserts that m(n,p,q)=2n−p−qm(n,p,q)=2^{n-p-q}. Equality is attained by the product family FP,Q={A⊆[n]:P⊆A and A∩Q=∅}\mathcal{F}_{P,Q}=\{A\subseteq[n]:P\subseteq A\text{ and }A\cap Q=\varnothing\}, where P,Q⊆[n]P,Q\subseteq[n] are disjoint with ∣P∣=p|P|=p and ∣Q∣=q|Q|=q.

References

Progress summary

Refreshed
Open · progress reported

A new unrefereed preprint proves the conjecture for many large parameter values, but the full finite-parameter problem remains open.

Frankl’s conjecture asserts that the natural product construction is optimal for the intersection–union problem. The reported result settles a broad range but not every finite choice of parameters.

October 2026 preprint

Fan Chang, Peter Frankl, and Jian Wang claim optimality when p,q≥2p,q \ge 2 and nn exceeds an explicit threshold, complementing earlier results for p=1p=1 or q=1q=1. This is an unrefereed preprint and does not cover all finite parameter values.

Current status (as of October 2026): The conjecture is claimed in the reported large-nn ranges, while the remaining finite parameter cases are open and the new result is unverified.

Sources

Solutions 0

No solutions have been posted yet.