← All problems

Erdős Problem #123

Erdős problem #123 · erdosproblems.com/123

Let a,b,c>1a,b,c>1 be pairwise coprime integers. Is every large integer a sum of distinct numbers of the form akblcma^k b^l c^m (k,l,m0k,l,m\ge 0), none dividing another?

Result
Proved
Field
Number Theory
Posed by
Paul Erdős, Mordechai Lewin
Year posed
1996
Years open
30y
Solved
2026-07
Model
GPT-5.6
Vendor
OpenAI
Collaborators
Colin Snyder
Verification
Lean-verified
Notability
No dedicated article

What the AI did

GPT-5.6 (prompted by Colin Snyder) resolved the Erdős-Lewin conjecture in the affirmative.

Verification

Marked proved (Lean) on erdosproblems.com; carried an Erdős prize of USD 250. Formally verified in Lean.

Source

erdosproblems.com