Erdős Problem #14
Editorial depth: Reviewed explanatory article — evidence qualified.
Snapshot classification: open (not automatically the live result).
Subject areas: number theory, sidon sets, 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
Given a set \(A\) of nonnegative integers, how often can an integer \(n\) have exactly one representation as a sum \(a+b\) with \(a,b\in A\)? The pair is unordered, so \(a+b\) and \(b+a\) are the same representation, and repeated summands are allowed. Write \(r_A(n)\) for the number of such pairs and
Erdős #14 asks two asymptotic questions about how few exceptions are possible, involving lower bounds near \(N^{1/2}\) and whether some single set has \(F_A(N)=o(\sqrt N)\). Both remain open in the inspected record. The two possibilities are not logically contradictory: a function like \(\sqrt N/\log(N+2)\) is simultaneously smaller than \(\sqrt N\) in relative terms and larger than \(N^{1/2-\varepsilon}\) up to constants for any fixed positive \(\varepsilon\).
A small example
For \(A=\{1,2,4\}\), the unordered pair sums with repetition are
Each listed integer has exactly one representation; 7 has none and is therefore an exception. In general the exceptions include both missing sums and sums with two or more different representations.
What is known
Representations can be computed by a finite convolution. Let \(x_j\) be the indicator of \(j\in A\), \(c_A(n)=\sum_{j=0}^n x_jx_{n-j}\) count ordered pairs, and let \(d_A(n)=1\) if \(n\) is even and \(n/2\in A\) (zero otherwise). Then
Non-diagonal unordered pairs contribute twice to \(c_A(n)\), while diagonal pairs contribute once. This gives an exact finite optimization and verification interface, not an asymptotic solution.
The literature includes older Erdős–Freud investigations of Sidon sets and unique sums. Subsequent finite computational reports must be distinguished from a construction of one infinite set that meets an asymptotic requirement at every sufficiently large cutoff. Optimizers for separate finite cutoffs need not fit together.
GCL's current state
Source audit #1283 checks unordered-pair and diagonal semantics against the pinned formal helper, supplies exact finite convolution and prefix-restriction identities, and records primary-source attribution limits. It also proves that shifting a nonnegative set by one translates pair-sum representations with a controlled finite change in the exception count. These are elementary interfaces, not MATHCERT proofs of either open assertion.
What remains
One direction requires a genuine universal lower bound for exceptions with its quantifiers and dependence on \(A\) stated correctly. The other seeks one fixed infinite set with exception count little-oh of \(\sqrt N\). Finite searches and convolution identities alone supply neither.
Sources and claim boundaries
- Erdős Problems #14: two-part question and maintained status.
- Erdős (1992), Some of my forgotten problems in number theory, Hardy–Ramanujan Journal 15, 34–50; page-level claims in the audit have not all been independently rechecked.
- Erdős–Freud (1991), On sums of a Sidon-sequence, Journal of Number Theory 38, 196–205. Do not misattribute different Hungarian source constructions to this paper's abstract.
- GCL source audit #1283: definitions and finite semantics, not a solution.
Catalogue and evidence provenance
- Pinned catalogue source: teorth/erdosproblems at
536408716724. - GCL issue: #1283 — protected issue/assignment identity; its labels and results are live, not reproduced automatically here.
- GCL triage: PROTECTED_READY_BACKLOG.
- Formal-statement coverage: A provider statement was included in the pinned snapshot. Presence does not mean the statement or proof has been checked.
- External mathematical status: consult current problem page and its dated bibliography.