Ruzsa’s genus-one problem

Fix a modulus N and look for a large set A ⊆ ℤ/Nℤ containing no nontrivial solutions to

a + 3b ≡ 2c + 2d (mod N)

with a, b, c, d ∈ A. A solution is trivial when a = b = c = d.

In a 1993 paper, Imre Z. Ruzsa suggested that |A| = N1−o(1) could be achievable. Our challenge: find an explicit witness with |A| > √N.

October 2026 update: Hametner and Solymosi proved that |A| ≥ N0.5608−o(1) is achievable, but their construction only beats √N for N much larger than 50,000.

2101​102​103​104​1101​102​ |A| = √N modulus N → record witness size |A| →

Each dot is the largest known witness for its modulus. A dot above the dashed line beats √N.

Browse witnesses →  ·  Leaderboard →  ·  Recent activity →  ·  Discuss on Zulip

Download all records (JSON) ↓

Submit a witness

Verification runs server-side in O(|A|²) time. Limits: N ≤ 50,000, |A| ≤ 10,000.

API

POST /api/verify with JSON body {"N": 25045, "A": [0, 260, ...]} returns the same verdict as JSON. See the API docs.