The Order of Long Rainbow Arithmetic Progressions
Let be the least such that every equinumerous -coloring of contains a rainbow -term arithmetic progression. Jungic, Licht, Mahdian, Nesetril and Radoicic conjectured ; Conlon, Fox and Sudakov proved . The matching lower bound holds, so and the conjectured order is wrong.
- Result
- Disproved(see note)
- Status
- Resolved
- AI contribution
- AI-assisted
- Method
- Construction
- Field
- Ramsey theory
- Posed by
- Veselin Jungic, Jacob Licht, Mohammad Mahdian, Jaroslav Nesetril, Rados Radoicic
- Year posed
- 2003
- Years open
- 23y
- Solved
- 2026-07-16
- Model
- Codex (GPT-5.6), Claude Code (Fable 5)
- Vendor
- OpenAI / Anthropic
- Collaborators
- Jesse Geneson
- Verification
- Unreviewed
- Publication
- Preprint
- Significance
- 20 / 100
- Disclosed cost
- —
- Wikipedia
- No dedicated article
What was actually shown
the true order is determined, and it is not the conjectured one
What the AI did
The acknowledgement says the two systems were used for proof exploration, proof criticism, exposition and revision. Proof exploration and criticism are mathematical work rather than prose work, but no specific step is attributed, so the lowest tier applies.
Verification
Single-author arXiv preprint; not yet peer-reviewed.
Source
arXiv:2607.15116 - The order of long rainbow arithmetic progressions