Erdős Problem #186
Editorial depth: Reviewed explanatory article — evidence qualified.
Snapshot classification: solved (not automatically the live result).
Subject areas: additive combinatorics.
Read this status correctly
The status above is reproduced from GCL's immutable external-catalogue snapshot. It is not a new claim of current mathematical resolution, a source-semantic review, an independent proof check, or GCL certification. A published proof and a GCL replay are separate events.
The question in plain language
A subset \(A\) of \(\{1,\ldots,N\}\) is non-averaging if no member is the arithmetic mean of two or more other distinct members. Let \(F(N)\) be the largest possible size of such a set. Originally asked by Straus and reported by Erdős, the problem is to determine the growth of \(F(N)\). It is more restrictive than avoiding three-term arithmetic progressions: averages of three or more elements must also be excluded.
A small example
The set \(\{1,2,4\}\) is non-averaging. The set \(\{1,2,3\}\) fails because \(2=(1+3)/2\). To check a candidate, choose \(a\in A\) and a subset \(B\subseteq A\setminus\{a\}\) with \(|B|\ge2\), and reject precisely when \(\sum_{b\in B}b=|B|a\). The members of \(B\) must be distinct; repetition would be a different condition.
What is known
SOLVED at the level of the growth exponent. Bosznay's 1989 construction gives \(F(N)\gg N^{1/4}\). Pham and Zakharov's Sharp bound for the Erdős–Straus non-averaging set problem proves \(F(N)\le N^{1/4+o(1)}\), building on work of Conlon, Fox and Pham. Therefore \(\log F(N)/\log N\to1/4\). This does not supply a leading constant or an exact formula. Erdős and Sárközy's earlier upper bound was \(O((N\log N)^{1/2})\).
The related #131 non-dividing condition is strictly stronger. It forbids a sum divisible by a selected member; non-averaging only forbids a sum equal to the number of summands times that member. The exact extremal functions should not be conflated.
GCL's current state
The pinned and maintained Erdős #186, updated April 2026, label the growth-order question SOLVED. Original source issue: #1455. The primary 2024 result's exponent was checked, but its full proof was not independently reconstructed.
The original issue contains an authenticated RESULT/1 source report. This is unadjudicated source evidence only, not independent mathematical certification or editorial admission.
What remains
A bounded successor enumerates every subset of \(\{1,\ldots,12\}\), checks all averages of two or more distinct other members using integer arithmetic, and computes the maximum size \(F(N)\) for small \(N\). Acceptance: \(\{1,2,3\}\) fails, \(\{1,2,4\}\) passes, and every rejection includes the witness \(a,B\). Stop at finite checks; do not claim a new asymptotic constant.
Sources and claim boundaries
- Pham–Zakharov, arXiv:2410.14624, primary sharp-exponent result.
- Erdős #186, Straus, Bosznay and earlier bounds.
- Original GCL source issue #1455.
No new extremal theorem, independent proof replay or editorial admission is claimed.
Catalogue and evidence provenance
- Pinned catalogue source: teorth/erdosproblems at
536408716724. - GCL issue: #1455 — protected issue/assignment identity; its labels and results are live, not reproduced automatically here.
- GCL triage: OUTSIDE_STRICT_OPEN_TRIAGE.
- Formal-statement coverage: No statement was confirmed in GCL's pinned formal-provider snapshot. Presence does not mean the statement or proof has been checked.
- External mathematical status: consult current problem page and its dated bibliography.